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
ILLUMINATE was a randomized, double-masked, parallel-group phase 3 treatment trial evaluating torcetrapib/atorvastatin versus atorvastatin in patients with coronary disease and diabetes mellitus. The registered primary endpoint was the time to first occurrence of a major cardiovascular disease event.
| Feature | ILLUMINATE |
|---|---|
| Phase | Phase 3 |
| Conditions | Coronary Disease; Diabetes Mellitus |
| Design | Randomized, parallel, double-masked |
| Primary purpose | Treatment |
| Enrollment | 15,067 |
| Interventions | Torcetrapib/atorvastatin; atorvastatin |
| Lead sponsor | Pfizer |
| Sponsor type | Industry |
| Status | Terminated |
| Start date | 2004-07 |
| Primary completion date | 2007-06 |
| ClinicalTrials.gov | NCT00134264 |
2. Clinical Question
The registered clinical question can be expressed as a randomized comparison of torcetrapib/atorvastatin with atorvastatin in patients with coronary disease and diabetes mellitus, using time to first major cardiovascular disease event as the primary endpoint.
Population
Patients with coronary disease and diabetes mellitus enrolled in a phase 3 treatment trial.
Intervention
Torcetrapib/atorvastatin.
Comparator
Atorvastatin.
Primary question
How does the time to first occurrence of a major cardiovascular disease event compare between the randomized treatment groups?
3. Trial Design
Torcetrapib/atorvastatin
- Randomized treatment intervention.
- Double-masked trial design.
- Primary endpoint based on time to first major cardiovascular disease event.
Atorvastatin
- Randomized comparator intervention.
- Double-masked trial design.
- Primary endpoint based on time to first major cardiovascular disease event.
4. Trial Timeline
Trial start
The registered study start date was July 2004.
Randomized treatment evaluation
The study used randomized allocation, a parallel design, double masking, and a treatment-focused primary purpose.
Primary completion
The registered primary completion date was June 2007.
Registry status
The ClinicalTrials.gov record identifies the study status as terminated.
5. Primary Endpoint
| Endpoint | Registry definition | Statistical structure |
|---|---|---|
| Major cardiovascular disease event | The time to first occurrence of a major cardiovascular disease event | Time-to-first-event endpoint |
The primary endpoint is a time-to-event outcome rather than a simple binary outcome assessed at a single fixed visit. Each participant can contribute information until the first qualifying event or until the participant is censored according to the prespecified follow-up rules.
The distinction between time to first event and the occurrence of any event during follow-up is important. A participant who experiences the first qualifying cardiovascular event early contributes a different amount of follow-up information from a participant who remains event-free for a much longer period.
6. Planned Analysis
The ClinicalTrials.gov record identifies the primary endpoint as the time to first occurrence of a major cardiovascular disease event, but it does not report posted statistical analyses for that endpoint.
For a randomized trial with a time-to-first-event primary endpoint, the usual statistical framework would begin with estimation of the event-time distribution in each randomized group. Kaplan-Meier methods are commonly used to estimate the probability of remaining free of the event over time, while a log-rank test is commonly used to compare the event-time distributions between groups.
A Cox proportional-hazards model is another standard approach for estimating a relative treatment effect through a hazard ratio. If used, the hazard ratio would describe the relative instantaneous event rate between randomized groups under the model. Its interpretation would depend on the model assumptions, particularly the proportional-hazards assumption.
The specific statistical analysis actually used for ILLUMINATE cannot be identified from the posted registry information alone.
7. Statistical Methodology
Time-to-first-event analysis
The primary endpoint records when the first qualifying major cardiovascular disease event occurs. This makes the endpoint naturally suited to survival-analysis methods, even though the clinical outcome is cardiovascular rather than mortality itself.
Time-to-event analysis is valuable because it uses both whether an event occurred and when the event occurred. A simple proportion of participants with events at the end of a study would discard much of that timing information.
Kaplan-Meier estimation
A Kaplan-Meier analysis estimates the probability of remaining free of the primary event as follow-up progresses. Participants who have not experienced the event by their last observed follow-up can contribute information up to that point and are then treated as censored under the analysis rules.
Here, di is the number of events at time ti and ni is the number at risk immediately before that time.
Hazard ratio
If a Cox model is used, the hazard ratio provides a relative measure of the instantaneous event rate between the randomized treatment groups. A value below 1 would indicate a lower estimated instantaneous event rate in the numerator treatment group under the fitted model; a value above 1 would indicate a higher estimated rate.
The hazard ratio is not the same as a risk ratio, an odds ratio, an absolute risk difference, or the percentage of patients who experience benefit.
Log-rank comparison
A log-rank test is commonly used to compare survival or event-free curves between randomized groups. It evaluates whether the observed pattern of event times differs between groups over follow-up, rather than comparing only one fixed time point.
Confidence intervals
A confidence interval around a hazard ratio quantifies statistical uncertainty around the estimated relative effect under the specified model and sampling framework. It does not describe the range of effects that individual participants experience.
8. Statistical Methods Explained
Why is the primary endpoint a time-to-event endpoint?
The registry defines the primary endpoint as the time to first occurrence of a major cardiovascular disease event. This preserves information about when events occur, rather than reducing follow-up to a single yes-or-no outcome.
Why would Kaplan-Meier estimation be appropriate?
Participants in a clinical trial can have different lengths of observable follow-up. Some may experience the endpoint, while others may remain event-free when their follow-up ends. Kaplan-Meier estimation is designed for this type of right-censored time-to-event data.
What would a hazard ratio describe?
A hazard ratio would compare the estimated instantaneous event rates between the two randomized groups. It summarizes relative event intensity over time under the assumptions of the fitted model. It would not directly state the absolute probability of experiencing a major cardiovascular disease event.
Why is the first event important?
The registry specifically defines the primary endpoint around the first occurrence of a major cardiovascular disease event. Once that first qualifying event occurs, the participant has reached the primary endpoint. Subsequent events would not change the time of the primary endpoint.
Why does censoring matter?
Participants who have not experienced the primary endpoint by the end of their observable follow-up can still provide useful information. Their data are handled through censoring rather than being treated as equivalent to participants who completed the same duration of follow-up without an event.
What would a p-value tell us?
If a formal hypothesis test were performed, the p-value would quantify the compatibility of the observed data with the specified null hypothesis under the statistical testing framework. It would not measure the size or clinical importance of an effect. Effect estimates and confidence intervals are needed to describe magnitude and precision.
9. Randomization and Masking
Participants were assigned using randomized allocation to the trial's treatment groups.
The treatment groups were evaluated as parallel randomized groups.
The registry identifies the study as double-masked.
The registry classifies the study's primary purpose as treatment.
Randomization and masking address different sources of bias. Randomization establishes the treatment comparison, while masking can reduce the influence of knowledge of treatment assignment on participant management, outcome assessment, or other trial processes.
10. Analysis Population and the ITT Principle
The registry identifies a randomized trial with 15,067 enrolled participants, but the available record does not provide a detailed definition of the analysis populations used for the primary endpoint.
For a randomized efficacy analysis, the intention-to-treat principle generally means retaining participants according to their randomized treatment assignment. This preserves the comparison created by randomization and avoids changing the treatment groups based on post-randomization events such as treatment discontinuation.
11. Multiplicity and Interim Analysis
The available ClinicalTrials.gov record identifies one registered primary endpoint: the time to first occurrence of a major cardiovascular disease event. It does not report a posted statistical-analysis description specifying multiplicity adjustments or an interim-analysis strategy.
Multiplicity becomes important when a trial evaluates multiple formal hypotheses, endpoints, treatment comparisons, or interim looks at the accumulating data. Without the prespecified testing strategy, a later numerical p-value cannot be interpreted correctly as either confirmatory or exploratory.
Multiple endpoints
If several endpoints are tested formally, the statistical plan needs to define how the overall type I error will be controlled.
Interim looks
If accumulating results are examined before the planned final analysis, the testing framework must account for repeated opportunities to reject the null hypothesis.
12. Missing Data and Censoring
For a time-to-first-event endpoint, missingness and censoring are not interchangeable concepts. A participant whose follow-up ends before the event may be censored, whereas an unrecorded endpoint assessment can create a missing-data problem requiring explicit handling.
The registry information available for ILLUMINATE does not report the specific censoring rules, missing-data strategy, or imputation method for the primary endpoint. Those details can materially affect a time-to-event analysis and are normally specified in the protocol or statistical analysis plan.
13. Non-Inferiority, Factorial, and Bayesian Methods
The registered design does not identify a non-inferiority framework. The available trial information also does not identify a factorial design or a Bayesian primary analysis.
| Design feature | What the registry identifies |
|---|---|
| Non-inferiority margin | Not reported in the available registry information |
| Factorial design | The registered design model is parallel |
| Bayesian primary analysis | Not reported in the available registry information |
| Interim-analysis method | Not reported in the available registry information |
| Multiplicity strategy | Not reported in the available registry information |
| Missing-data/imputation method | Not reported in the available registry information |
14. Understanding the Primary Endpoint Statistically
The endpoint is defined around the first occurrence of a major cardiovascular disease event. Statistically, that means the analysis concerns the time from the relevant starting point to the first qualifying event.
A hazard ratio, if used, would summarize the relative instantaneous rate of reaching the first event between randomized groups. It would not directly provide an absolute treatment effect at a particular time point.
Kaplan-Meier estimates at clinically meaningful time points can provide absolute event-free probabilities. Those estimates complement a hazard ratio because they describe the event-free experience on an absolute scale.
A confidence interval would show the uncertainty surrounding the estimated treatment effect. Narrower intervals generally indicate greater statistical precision, while wider intervals indicate greater uncertainty.
15. Limitations
- No posted statistical analyses: the ClinicalTrials.gov record does not provide numerical primary-endpoint estimates, confidence intervals, or p-values.
- Limited statistical-plan detail: the available registry information does not specify the formal survival-analysis model, censoring rules, multiplicity strategy, interim-analysis method, or missing-data approach.
- Endpoint definition: the primary endpoint is described as time to first occurrence of a major cardiovascular disease event, but the available record does not provide a more detailed component-level definition.
- Analysis population: the registry reports enrollment of 15,067 but does not provide the detailed analysis-population definitions needed to determine exactly which participants would enter each statistical analysis.
- Time-to-event assumptions: if a Cox proportional-hazards model were used, interpretation would depend on the proportional-hazards assumption. The available record does not document whether that assumption was assessed.
- Censoring: valid survival analysis depends on appropriate censoring rules and assumptions. Those rules are not specified in the available registry information.
- Terminated status: the study is listed as terminated. The registry information available here does not provide a statistical explanation for that status.
16. Why This Trial Matters Statistically
ILLUMINATE is a useful teaching case for understanding how a large randomized treatment trial can be structured around a cardiovascular time-to-event endpoint. The design combines randomization, masking, parallel treatment groups, and a primary endpoint that incorporates both whether and when the first major cardiovascular disease event occurs.
| Concept | How it appears in ILLUMINATE |
|---|---|
| Randomization | The allocation is registered as randomized. |
| Parallel-group design | The registered design model is parallel. |
| Double masking | The registry identifies the study as double-masked. |
| Time-to-event analysis | The primary endpoint is time to first major cardiovascular disease event. |
| Kaplan-Meier estimation | A standard approach for estimating event-free probability over time. |
| Hazard ratio | A standard relative measure for comparing event rates over follow-up if a Cox model is specified. |
| Censoring | A central consideration whenever follow-up times differ between participants. |
| Confidence intervals | Needed to describe precision around an estimated treatment effect. |
| Multiplicity | Requires attention if multiple formal hypotheses or interim analyses are performed. |
17. What a Complete Statistical Analysis Would Examine
A complete analysis of the registered primary endpoint would ordinarily connect the clinical endpoint definition to a prespecified statistical estimand and analysis population.
Step 1 · Define the event
Specify exactly which cardiovascular outcomes constitute a major cardiovascular disease event and which event counts as the first endpoint event.
Step 2 · Define time zero
Specify the starting point from which time to the first event is measured.
Step 3 · Handle censoring
Define when participants who do not experience the event are censored and what assumptions support that approach.
Step 4 · Estimate survival
Use a suitable time-to-event estimator, commonly Kaplan-Meier, to describe event-free probability over time.
Step 5 · Compare groups
Use a prespecified comparison such as a log-rank test and/or a model-based treatment-effect estimate.
Step 6 · Quantify uncertainty
Report confidence intervals alongside effect estimates so the magnitude and precision of the treatment comparison can be assessed.
18. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary endpoint is a time-to-event outcome for which treatment groups can be compared using methods that account for both event occurrence and follow-up time.
Clinical interpretation
The clinical meaning of the treatment comparison depends on the exact cardiovascular event definition, the magnitude and precision of the observed treatment effect, and the balance of outcomes and safety findings reported for the trial.
19. Sources
- ClinicalTrials.gov: NCT00134264 — ILLUMINATE.
Continue through the Clinical Biostats clinical-trial library
Clinical trial analyses can be read alongside statistical methods and study-design concepts to connect endpoint definitions with the models used to evaluate them.
20. Record Summary
ILLUMINATE was a phase 3 randomized, double-masked, parallel-group treatment trial enrolling 15,067 participants with coronary disease and diabetes mellitus. The intervention was torcetrapib/atorvastatin and the comparator was atorvastatin. Its registered primary endpoint was the time to first occurrence of a major cardiovascular disease event, making survival-analysis methods the natural statistical framework for evaluating the primary treatment comparison. The ClinicalTrials.gov record identifies the study as terminated, with a start date of 2004-07 and a primary completion date of 2007-06.