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
ACTIVE-W was a randomized, parallel, phase 3 prevention trial in atrial fibrillation and vascular risk. The registry reports an enrollment of 6706 and identifies clopidogrel (SR25990C) as the drug intervention. Its primary outcome was a time-to-event composite endpoint assessed during approximately three years of follow-up.
| Feature | ACTIVE-W |
|---|---|
| Phase | Phase 3 |
| Conditions | Atrial Fibrillation; Vascular Risk |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Prevention |
| Enrollment | 6706 |
| Lead sponsor | Sanofi |
| Sponsor type | Industry |
| Intervention listed in the registry | clopidogrel (SR25990C) (drug) |
| Status | Terminated |
| ClinicalTrials.gov | NCT00243178 |
2. Clinical Question
The registry identifies ACTIVE-W as a randomized phase 3 prevention trial in people with atrial fibrillation and vascular risk. Its principal statistical question is expressed through a time-to-first-event composite outcome: how long it takes after trial allocation for a participant to experience the first occurrence of one of the specified vascular events.
Population
Participants enrolled in a phase 3 prevention trial for atrial fibrillation and vascular risk.
Intervention
Clopidogrel (SR25990C), listed in the registry as a drug intervention.
Comparator
The ClinicalTrials.gov record identifies the randomized trial design but does not report a comparator intervention in the registry information summarized here.
Primary question
For the randomized comparison, how does treatment relate to time to the first occurrence of the prespecified composite vascular outcome?
3. Trial Design
The combination of randomization and a parallel design is important statistically. Randomization is intended to make treatment assignment independent of measured and unmeasured baseline prognostic factors on average, while the parallel structure means participants remain associated with their randomized group rather than being deliberately switched between randomized strategies as part of a crossover design.
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Primary outcome | Time to the first occurrence of stroke, non-CNS systemic embolism, myocardial infarction or vascular death | During approximately three years of follow-up |
The primary endpoint is a composite time-to-event endpoint. Two features are important statistically. First, the analysis concerns the time until an event occurs, not simply whether an event occurred at a fixed visit. Second, the endpoint counts the first occurrence of any of the four listed event types.
That first-event structure means that a participant who experiences one component of the composite becomes an event for the primary endpoint at that point. The primary analysis therefore concerns time to the first qualifying event rather than the total number of vascular events accumulated by each participant.
5. Statistical Methodology
The registry defines the primary outcome as time to the first occurrence of a composite event during approximately three years of follow-up. Endpoints of this form are ordinarily analyzed with time-to-event methods, because participants can have different follow-up times and some participants may remain event-free when observation ends.
Kaplan-Meier estimation
A Kaplan-Meier estimator can describe the probability of remaining free of the first composite event over time. It accounts for right censoring by allowing a participant who has not experienced the endpoint by the last observation to contribute information up to that time.
Here, di represents the number of first events at event time ti, while ni represents the number of participants at risk immediately before that time.
Hazard ratio
A Cox proportional-hazards model is a standard way to compare time to the first event between randomized groups. Its treatment coefficient can be expressed as a hazard ratio:
For a treatment indicator, the hazard ratio compares the estimated instantaneous event rates between the treatment groups, conditional on the fitted model.
A hazard ratio below 1 would indicate a lower estimated instantaneous rate of the first composite event in the intervention group relative to the comparator under the model. A hazard ratio above 1 would indicate a higher estimated instantaneous rate. A hazard ratio of exactly 1 would indicate no estimated relative difference in hazard.
Log-rank comparison
A log-rank test is commonly paired with Kaplan-Meier estimation when comparing two randomized groups over time. Rather than comparing only one fixed time point, it uses the ordering of observed event times throughout follow-up.
Confidence intervals
A confidence interval around a hazard ratio quantifies statistical uncertainty in the estimated treatment effect. A narrow interval indicates greater statistical precision than a wide interval, all else being equal. The interval does not describe the range of effects that must occur for individual participants.
Analysis population
Because the registry information summarized here identifies randomization but does not provide a detailed statistical analysis plan, the precise analysis population and handling of treatment discontinuation are not reported. In a randomized efficacy analysis, an intention-to-treat framework would generally preserve the original randomized comparison.
6. Statistical Methods Explained
Why is this a time-to-event problem?
The endpoint is explicitly defined as time to the first occurrence of stroke, non-CNS systemic embolism, myocardial infarction, or vascular death. A participant who remains event-free throughout observation contributes follow-up information even without an observed event. This makes ordinary proportions alone incomplete for describing the primary endpoint.
Why use a composite endpoint?
A composite endpoint combines several clinically specified event types into one analysis target. Statistically, this can increase the number of observed first events and therefore provide more information about the overall event process. The interpretation, however, is specifically about the first occurrence of any component, not necessarily about each component individually.
What does "first occurrence" mean?
The endpoint is triggered by the earliest qualifying event. If a participant experiences more than one component during follow-up, the primary time-to-event analysis treats the first qualifying occurrence as the endpoint event. This prevents repeated events from being counted multiple times in the primary first-event analysis.
Why can Kaplan-Meier analysis be preferable to a simple event percentage?
A simple percentage can ignore differences in observation time. Kaplan-Meier analysis incorporates the timing of events and permits participants with incomplete event observation to be censored rather than automatically treated as if they remained event-free indefinitely.
What does a hazard ratio measure?
The hazard ratio is a relative comparison of instantaneous event rates under the fitted survival model. It is not a ratio of cumulative probabilities at a particular time point, and it is not the percentage of participants who benefit.
Why does censoring matter?
Participants can reach the end of their observed follow-up without experiencing the composite endpoint, or their observation can otherwise end before an event occurs. Survival methods treat such observations as censored at their last usable follow-up time. The validity of the resulting estimates depends on assumptions about the censoring mechanism.
What is the proportional-hazards assumption?
The conventional Cox proportional-hazards interpretation assumes that the relative hazard between groups is reasonably stable over time. If the relative effect changes substantially during follow-up, a single hazard ratio may provide an incomplete description, and time-specific survival estimates or other models may be needed.
7. Planned Analysis
The ClinicalTrials.gov record defines the primary outcome as time to the first occurrence of stroke, non-CNS systemic embolism, myocardial infarction or vascular death during approximately three years of follow-up.
For an endpoint of this type, the statistical analysis would ordinarily include:
- Kaplan-Meier estimation of the time-to-first-event distribution in the randomized groups.
- A comparison of the event-time distributions using a log-rank framework.
- Estimation of a relative treatment effect using a Cox proportional-hazards model, commonly expressed as a hazard ratio with a confidence interval.
- Assessment of the number and timing of first composite events in each randomized group.
- Consideration of censoring for participants who are event-free when their follow-up ends.
8. Interpreting the Primary Endpoint
The most important statistical feature of ACTIVE-W is that the primary outcome combines both event type and event timing. This has several consequences for interpretation.
What is measured
The time from the relevant trial starting point to the first occurrence of stroke, non-CNS systemic embolism, myocardial infarction, or vascular death.
What is not measured
The primary endpoint is not the total number of vascular events experienced by each participant.
Why timing matters
An event occurring early and an event occurring late are both endpoint events, but time-to-event methods retain information about when they occurred.
Why the composite matters
The primary result would represent the combined first-event process across all four prespecified component outcomes.
9. Composite Endpoint Interpretation
Composite endpoints can be statistically efficient because several clinically important outcomes contribute to one primary event definition. But the components can differ in frequency, severity, clinical meaning, and treatment responsiveness.
For ACTIVE-W, the composite consists of stroke, non-CNS systemic embolism, myocardial infarction, and vascular death. A treatment effect on the composite therefore should not automatically be interpreted as an identical effect on each individual component.
The primary endpoint is triggered by the earliest qualifying component. Component-specific analyses would answer different questions and would require their own reported event counts and statistical estimates.
The registry information summarized here does not report component-specific event counts or component-specific treatment estimates. Consequently, the primary endpoint should be understood as the combined first-event outcome rather than as evidence about the relative frequency of any one component.
10. Randomization and Causal Interpretation
Randomization is central to the interpretation of a clinical trial. In a randomized comparison, treatment assignment is determined by the trial design rather than by participants' prognostic characteristics. With adequate implementation, this reduces systematic differences between groups at baseline and supports causal interpretation of differences in outcomes between randomized groups.
ACTIVE-W is explicitly identified as randomized and parallel. The registry also identifies the primary purpose as prevention. These features support a framework in which the treatment comparison is evaluated through differences in subsequent clinical event experience.
Randomization
Creates the basis for comparing outcomes according to assigned treatment rather than according to self-selected treatment.
Parallel design
Participants remain in their assigned randomized treatment pathway rather than being deliberately crossed between study strategies as part of the design.
No masking
The registry identifies the trial as having no masking. This is a design characteristic that can matter for outcome assessment and treatment behavior, depending on the endpoint and study procedures.
Prevention objective
The registry identifies prevention as the trial's primary purpose, placing the composite vascular endpoint in a preventive clinical context.
11. Censoring and Follow-Up
The approximately three-year follow-up window means that the primary endpoint is evaluated over time rather than at a single fixed assessment. Participants who have not experienced a qualifying event at the end of their available observation can be censored in a conventional time-to-event analysis.
Censoring is not equivalent to an event. A censored participant contributes information about being event-free through the observed follow-up period, but the analysis does not treat that participant as known to have remained event-free indefinitely.
The quality of a survival analysis depends partly on whether the censoring process is appropriately handled. If censoring is related to prognosis in ways not adequately addressed by the analysis, estimated event-free survival and treatment comparisons can be affected.
12. Hazard Ratios: A Practical Interpretation
A hazard ratio below 1 would indicate a lower estimated instantaneous rate of experiencing the first composite endpoint in the clopidogrel group relative to the comparator, under the fitted model.
A hazard ratio above 1 would indicate a higher estimated instantaneous rate of the first composite endpoint in the clopidogrel group relative to the comparator, under the fitted model.
A hazard ratio of 1 would correspond to no estimated relative difference in instantaneous event rate between the randomized groups under the model.
These interpretations concern a relative rate over time. They should not be converted directly into an absolute reduction in the probability of an event. Absolute risk depends on the underlying event rate and the duration of follow-up.
13. Confidence Intervals and P-Values
The registry record summarized here does not post a confidence interval or p-value for the primary endpoint. In a completed statistical analysis, these quantities would normally accompany the treatment-effect estimate.
| Statistic | What it would address | What it would not establish by itself |
|---|---|---|
| Hazard ratio | Relative difference in instantaneous event rates under the fitted survival model | Absolute benefit, individual patient benefit, or causal effect on every component separately |
| 95% confidence interval | Statistical uncertainty around the estimated treatment effect | The range of outcomes that individual participants will experience |
| P-value | Evidence against a specified null hypothesis under the statistical model | The size, clinical importance, or probability that the treatment hypothesis is true |
This distinction is particularly important for time-to-event trials. Statistical evidence and clinical magnitude are related but different questions. A complete interpretation would normally examine the effect estimate, its uncertainty, the absolute event experience over time, and the clinical importance of the endpoint.
14. Multiplicity and Composite Components
The registry information summarized here identifies one registered primary composite endpoint. It does not report a multiplicity strategy for additional endpoints, component analyses, subgroup analyses, or interim looks.
If multiple statistical hypotheses are tested in a full analysis, the interpretation of nominal p-values depends on how those tests relate to the prespecified hypothesis family. Testing many secondary or component outcomes without appropriate error control can increase the chance of observing a statistically unusual result by chance alone.
15. Interim Analysis, Stratification, and Bayesian Methods
The registry information summarized here does not report an interim-analysis plan, alpha-spending procedure, stratification factors, or Bayesian statistical methods.
| Design topic | Registry status in the record summarized here |
|---|---|
| Interim analysis | Not reported |
| Alpha-spending method | Not reported |
| Stratification factors | Not reported |
| Non-inferiority margin | Not reported |
| Factorial design | Not reported |
| Crossover design | Not reported |
| Missing-data / imputation method | Not reported |
| Bayesian methods | Not reported |
These omissions matter because each can change how an eventual statistical analysis should be interpreted. For example, an interim efficacy analysis requires appropriate control of the overall type I error, while stratification affects the specification of a stratified analysis. Neither feature should be assumed without documentation.
16. Safety and Secondary Outcomes
The registry information summarized here identifies the primary outcome but does not report secondary endpoint results or serious adverse event counts by randomized arm.
Safety and efficacy answer different statistical questions. A time-to-first vascular endpoint describes the occurrence of a prespecified clinical outcome, whereas adverse-event analyses generally describe treatment-emergent events, their severity, seriousness, and relationship to exposure.
Efficacy
The registered primary efficacy outcome is time to the first occurrence of the composite vascular endpoint.
Safety
The registry information summarized here does not provide serious adverse event counts by arm.
Secondary endpoints
No secondary endpoint results are reported in the information summarized here.
Component outcomes
No component-specific event counts or effect estimates are reported in the information summarized here.
17. Trial Timeline
Trial start
The registry lists July 2003 as the trial start date.
Randomized prevention study
The trial is registered as phase 3 with randomized allocation, a parallel design, no masking, and prevention as its primary purpose.
Primary completion
The registry lists September 2005 as the primary completion date.
Terminated
The ClinicalTrials.gov record identifies the study status as terminated.
18. What the Primary Endpoint Does — and Does Not — Mean
The endpoint measures the time until the first occurrence of stroke, non-CNS systemic embolism, myocardial infarction, or vascular death during approximately three years of follow-up.
It does not directly measure the total number of vascular events per participant, the severity of every event, or the separate treatment effect on every component.
The four component outcomes represent different manifestations of vascular disease. Combining them creates a single primary event definition, but the statistical result would still require clinical interpretation of which components contributed to the observed difference.
19. Limitations and Interpretation Issues
- No posted statistical results: the ClinicalTrials.gov record summarized here does not report an effect estimate, confidence interval, or p-value for the primary endpoint.
- Comparator information: the registry information summarized here identifies clopidogrel as the intervention but does not report a comparator intervention.
- Composite endpoint: the primary outcome combines four different clinical events, so a composite treatment effect would not automatically establish the same effect for every component.
- First-event analysis: the endpoint concerns the first qualifying occurrence, not recurrent vascular events.
- Time-to-event assumptions: conventional Cox modeling relies on assumptions concerning the relationship between hazards over time.
- Censoring: valid survival inference depends on appropriate handling and assumptions concerning censored observations.
- Unreported analysis details: the registry information summarized here does not specify stratification, interim monitoring, missing-data procedures, multiplicity adjustments, or Bayesian methods.
- No component results: without event counts and estimates for individual components, the relative contribution of stroke, systemic embolism, myocardial infarction, and vascular death cannot be quantified from the registry record summarized here.
20. Why This Trial Matters Statistically
ACTIVE-W provides a useful teaching example of how randomized prevention trials can be structured around clinically meaningful time-to-event outcomes. Its primary endpoint is not simply whether an event happened; it is when the first qualifying event happened, with the event defined as the first occurrence of any of four vascular outcomes.
| Statistical concept | How it appears in ACTIVE-W |
|---|---|
| Randomization | The registry identifies randomized allocation. |
| Parallel design | The study is registered as a parallel trial. |
| Time-to-event analysis | The primary endpoint is explicitly defined as time to first occurrence. |
| Composite endpoint | Four vascular outcomes form the primary composite. |
| First-event analysis | The endpoint is triggered by the first occurrence of a qualifying component. |
| Kaplan-Meier estimation | A standard framework for describing event-free time with censoring. |
| Cox regression | A standard model for estimating a relative time-to-event effect. |
| Log-rank testing | A standard group comparison for time-to-event distributions. |
| Censoring | Relevant when participants remain event-free through their available follow-up. |
| Clinical interpretation | Composite results require attention to the individual component outcomes. |
21. Statistical Concepts in This Trial
The most useful statistical lessons from ACTIVE-W are the distinction between a binary outcome and a time-to-event outcome, the construction of composite endpoints, and the interpretation of hazard ratios in randomized clinical trials.
22. Sources
- ClinicalTrials.gov: ACTIVE-W, NCT00243178.
- PubMed: PubMed record.
- PubMed: PubMed record.
- PubMed: PubMed record.
- PubMed: PubMed record.
Continue through Clinical Biostats
Explore additional clinical trial results and statistical resources covering the methods used to design, analyze, and interpret randomized clinical trials.
23. Record Summary
ACTIVE-W is a randomized phase 3 prevention trial involving 6706 participants with atrial fibrillation and vascular risk. The registry identifies clopidogrel (SR25990C) as the drug intervention and defines the primary outcome as time to the first occurrence of stroke, non-CNS systemic embolism, myocardial infarction or vascular death during approximately three years of follow-up.
Statistically, the trial is most naturally understood as a time-to-event comparison. Kaplan-Meier methods can describe event-free time, log-rank methods can compare randomized groups over follow-up, and a Cox model can express the relative treatment effect as a hazard ratio with a confidence interval. The composite nature of the endpoint is equally important: the primary result concerns the first occurrence of any qualifying component, not the total burden of recurrent events or an automatically identical effect on each component.