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
dal-GenE was a randomized, parallel-group, quadruple-masked phase 3 cardiovascular trial evaluating dalcetrapib versus placebo in a genetically defined population with a recent acute coronary syndrome. The primary efficacy endpoint was a composite time-to-event measure of cardiovascular death, resuscitated cardiac arrest, non-fatal myocardial infarction, and non-fatal stroke.
| Feature | dal-GenE |
|---|---|
| Phase | Phase 3 |
| Condition | Acute Coronary Syndrome |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 6,147 |
| Arms | 2 |
| Interventions | Dalcetrapib and placebo |
| Start | April 2016 |
| Primary completion | October 2021 |
| Status | Completed |
| Lead sponsor | DalCor Pharmaceuticals |
2. Clinical Question
The primary statistical question was whether dalcetrapib changed the time to first occurrence of a composite cardiovascular endpoint compared with placebo in participants with a recent acute coronary syndrome who met the trial's genetically defined population criteria.
Population
A genetically defined population with a recent acute coronary syndrome.
Intervention
Dalcetrapib.
Comparator
Placebo.
Primary question
Does dalcetrapib alter the hazard of the composite cardiovascular endpoint relative to placebo?
3. Trial Design
Dalcetrapib
- Dalcetrapib
- Randomized treatment assignment
- Primary comparison against placebo
Placebo
- Placebo
- Randomized treatment assignment
- Reference group for the efficacy comparison
The quadruple-masked design is important statistically because knowledge of treatment assignment can influence participant behavior, clinical management, outcome assessment, and other aspects of trial conduct. The registry states that all efficacy endpoints were adjudicated by an independent clinical endpoint committee.
Trial timeline
| Milestone | Date |
|---|---|
| Study start | April 2016 |
| Primary completion | October 2021 |
| Status | Completed |
4. Endpoints
The primary endpoint and the two posted secondary endpoints were all time-to-event composites. The registry reports a median duration of follow-up of 39.9 months for these analyses.
| Endpoint | Registry definition / time frame | Statistical role |
|---|---|---|
| Composite of Cardiovascular Death, Resuscitated Cardiac Arrest, Non-Fatal Myocardial Infarction, and Non-Fatal Stroke | From randomization to the first occurrence of any component of the composite primary endpoint (median duration of follow-up was 39.9 months). All efficacy endpoints were adjudicated by an independent clinical endpoint committee. | Primary |
| Composite Endpoint of Cardiovascular Death, Resuscitated Cardiac Arrest, Non-Fatal Myocardial Infarction, Non-Fatal Stroke, Hospitalization for ACS (With Electrocardiogram Abnormalities) or Unanticipated Coronary Revascularization | From randomization to the first occurrence of any component of the composite secondary endpoint (median duration of follow-up was 39.9 months). | Secondary |
| Composite Endpoint of Cardiovascular Death, Resuscitated Cardiac Arrest, Non-Fatal Myocardial Infarction, Non-Fatal Stroke or Hospitalization for New or Worsening Heart Failure | From randomization to the first occurrence of any component of the composite secondary endpoint (median duration of follow-up was 39.9 months). | Secondary |
The primary endpoint is a time-to-first-event composite. A participant contributes the time from randomization until the first positively adjudicated component of the composite. Once the first qualifying component occurs, subsequent components do not create additional primary endpoint events for that participant.
5. Statistical Methodology
Primary analysis: stratified Cox proportional-hazards model
The primary endpoint was analyzed using a stratified Cox proportional-hazards model. Treatment was included as a main effect, while region and acute coronary syndrome index event type were used as stratification factors.
Here, λ represents the assumed constant hazard ratio for time to occurrence of the composite primary endpoint comparing dalcetrapib with placebo.
The analysis population was all subjects randomized to treatment. This is important because randomization establishes the treatment groups before subsequent follow-up events, treatment discontinuation, or other post-randomization experiences occur.
Why stratification was used
Stratification allows the Cox model to account for specified factors without requiring a single common baseline hazard function across all strata. In dal-GenE, region and ACS index event type were the stated stratification factors for the primary model.
The treatment hazard ratio is therefore estimated from comparisons within the stratified survival-analysis framework rather than from an unadjusted comparison that ignores those factors.
Superiority framework
The registered hypothesis type was superiority. The null hypothesis specified a hazard ratio of 1, while the two-sided alternative allowed the treatment hazard ratio to differ from 1 in either direction.
Secondary endpoint analysis
The two secondary endpoints were also time-to-event outcomes analyzed using a stratified Cox proportional-hazards model. Their analyses were described as similar to the primary analysis.
6. Multiplicity and Secondary Endpoints
The registry explicitly describes a multiplicity strategy for the secondary endpoints. To control family-wise Type I error arising from multiple endpoints, the secondary endpoints were to be formally tested using Hochberg's step-up procedure only if the primary analysis demonstrated a significant treatment effect at p<0.05.
Primary endpoint first
The secondary endpoints were linked to the success criterion for the primary analysis. If the primary analysis did not produce a significant treatment effect at p<0.05, secondary statistical tests were presented for illustrative purposes rather than as formally controlled confirmatory tests.
Why this matters
Testing several hypotheses increases the opportunity for a false-positive finding. A prespecified multiplicity procedure provides a framework for controlling the family-wise Type I error rate.
This hierarchy changes how the secondary p-values should be read. A nominal p-value is not automatically equivalent to a formally confirmatory result when the prespecified testing sequence has not been satisfied.
7. Primary Result
Composite cardiovascular endpoint
Hazard ratio: Dalcetrapib vs Placebo
95% CI: 0.75–1.03 · P = 0.12
Median duration of follow-up: 39.9 months
| Analysis element | Reported result |
|---|---|
| Analysis population | All subjects randomized to treatment |
| Comparison | Placebo vs Dalcetrapib |
| Endpoint | Composite of cardiovascular death, resuscitated cardiac arrest, non-fatal myocardial infarction, and non-fatal stroke |
| Method | Stratified Cox proportional-hazards model |
| Effect measure | Hazard ratio |
| Estimate | 0.88 |
| 95% confidence interval | 0.75–1.03 |
| P-value | 0.12 |
| Hypothesis | Superiority |
| Stratification factors | Region and ACS index event type |
The estimated hazard ratio of 0.88 means that, under the fitted stratified Cox model, the estimated instantaneous rate of experiencing the first event in the composite endpoint was approximately 12% lower with dalcetrapib than with placebo. That is a relative hazard estimate; it is not a statement that 12% of participants avoided an event, nor does it mean that every participant experienced the same reduction in risk.
The 95% confidence interval of 0.75–1.03 describes the uncertainty around the estimated hazard ratio under the model and sampling framework. Because the interval extends above 1, values compatible with the data include a modest increase as well as a decrease in the estimated hazard.
The p-value of 0.12 describes the evidence against the specified null hypothesis of a hazard ratio equal to 1 under the statistical model. It does not measure the magnitude or clinical importance of the effect. Effect magnitude is conveyed by the hazard ratio and its confidence interval.
The interpretation also depends on the Cox-model framework, including its proportional-hazards assumption. A single hazard ratio is most straightforward to interpret when the relative hazards are reasonably stable over time. The endpoint is additionally a time-to-first-event composite, so the estimate does not describe the separate effect on each component.
8. What the Primary Hazard Ratio Means
The primary estimate can be expressed in a simple relative-hazard interpretation: an HR of 0.88 corresponds to an estimated hazard that is 88% of the placebo hazard under the fitted model, or approximately a 12% lower estimated hazard.
The 0.12 represents the relative reduction in estimated hazard, not an absolute risk reduction and not the percentage of participants who benefit.
It is also important not to translate the hazard ratio directly into a difference in cumulative incidence. Survival probabilities depend on the underlying event process and follow-up, whereas the Cox hazard ratio summarizes a relative instantaneous-event-rate comparison.
The confidence interval provides an important second layer of interpretation. An estimate of 0.88 is relatively close to the null value of 1, and the interval extends from 0.75 to 1.03. The interval therefore communicates substantially more information about uncertainty than the point estimate alone.
9. Secondary Endpoint Results
Secondary composite: ACS hospitalization or unanticipated coronary revascularization
Hazard ratio: Dalcetrapib vs Placebo
95% CI: 0.88–1.13 · P = 0.95
Median duration of follow-up: 39.9 months
| Element | Result |
|---|---|
| Endpoint | Composite of cardiovascular death, resuscitated cardiac arrest, non-fatal myocardial infarction, non-fatal stroke, hospitalization for ACS (with electrocardiogram abnormalities) or unanticipated coronary revascularization |
| Analysis population | All subjects randomized to treatment |
| Method | Stratified Cox proportional-hazards model |
| Hazard ratio | 1.00 |
| 95% CI | 0.88–1.13 |
| P-value | 0.95 |
An HR of 1.00 is exactly the null value for a hazard-ratio comparison. The estimate therefore indicates essentially no difference in the estimated hazard between the two randomized groups for this composite endpoint.
The 95% CI of 0.88–1.13 describes the uncertainty around that estimate. It includes 1 and therefore spans both modestly lower and modestly higher hazards.
The p-value of 0.95 is very close to the null-model probability calculation under the specified test. It should not be interpreted as the probability that the treatment has no effect, nor does it quantify clinical equivalence.
The registry specifies that secondary endpoints were formally tested under the Hochberg procedure only if the primary analysis was significant at p<0.05. Because the primary result had a reported p-value of 0.12, this secondary test was presented solely for illustrative purposes under the registered multiplicity strategy.
Secondary composite: heart failure
Hazard ratio: Dalcetrapib vs Placebo
95% CI: 0.79–1.07 · P = 0.27
Median duration of follow-up: 39.9 months
| Element | Result |
|---|---|
| Endpoint | Composite of cardiovascular death, resuscitated cardiac arrest, non-fatal myocardial infarction, non-fatal stroke or hospitalization for new or worsening heart failure |
| Analysis population | All subjects randomized to treatment |
| Method | Stratified Cox proportional-hazards model |
| Hazard ratio | 0.92 |
| 95% CI | 0.79–1.07 |
| P-value | 0.27 |
The estimated HR of 0.92 corresponds to an approximately 8% lower estimated hazard under the fitted model. The estimate itself is modest in magnitude and remains close to the null value of 1.
The 95% CI of 0.79–1.07 includes 1, so the uncertainty interval is compatible with both a lower and a higher hazard. The interval is more informative than the point estimate alone because it shows the range of treatment effects reasonably compatible with the model and observed data.
The p-value of 0.27 does not measure the size of the estimated treatment effect. It describes the evidence against the null hypothesis within the specified testing framework.
As with the other secondary endpoint, the registry states that formal multiplicity-controlled testing depended on a significant primary analysis. The reported primary p-value was 0.12, so the secondary statistical test was presented for illustrative purposes under the registered hierarchy.
10. Results in Context: Three Hazard Ratios
| Endpoint | HR | 95% CI | P-value | Role |
|---|---|---|---|---|
| Primary composite | 0.88 | 0.75–1.03 | 0.12 | Primary |
| Secondary composite with ACS hospitalization / revascularization | 1.00 | 0.88–1.13 | 0.95 | Secondary |
| Secondary composite with new or worsening heart failure | 0.92 | 0.79–1.07 | 0.27 | Secondary |
All three posted analyses used the same broad statistical framework: randomized participants, time-to-event outcomes, and stratified Cox proportional-hazards modeling. The differences in the endpoint definitions are therefore important when interpreting the estimates; the three hazard ratios do not describe the same clinical event.
The primary estimate of 0.88 is lower than 1, whereas the first secondary estimate is exactly 1.00 and the second secondary estimate is 0.92. Their confidence intervals overlap the null value of 1. This pattern should be read as a set of endpoint-specific estimates rather than as a single overall treatment effect.
11. Statistical Methods Explained
Why was a Cox proportional-hazards model used?
The primary and secondary endpoints measure the time from randomization to the first occurrence of a specified event. Time-to-event methods are designed for this setting because they can incorporate different follow-up times and censoring. The Cox model then provides a relative hazard estimate comparing treatment groups.
What does an HR of 0.88 mean?
An HR of 0.88 means that the estimated instantaneous event rate under the model was 88% of the corresponding rate in the comparator group. Expressed as a relative difference, this is approximately a 12% lower estimated hazard. It is not equivalent to saying that 12% fewer participants had an event.
Why was the model stratified?
The primary model used region and ACS index event type as stratification factors. Stratification allows the analysis to account for those factors while estimating the treatment effect across the stratified survival-analysis framework.
Why does the confidence interval matter?
The point estimate is only one estimate of the treatment effect. The 95% confidence interval shows the uncertainty around that estimate under the statistical model. For the primary endpoint, the interval from 0.75 to 1.03 indicates that the data are compatible with a range of possible hazard ratios around the point estimate.
Why does the p-value not measure effect size?
A p-value evaluates the compatibility of the observed data with a specified null hypothesis under the chosen statistical model. It does not describe how large or clinically important the treatment effect is. Effect magnitude is better described by the hazard ratio, confidence interval, and—when available—absolute event probabilities or other clinically interpretable measures.
Why does multiplicity matter for the secondary endpoints?
Multiple endpoint tests create a larger opportunity for at least one false-positive result. The registry therefore specified a Hochberg step-up procedure for formal testing of the secondary endpoints, conditional on a significant primary analysis. Because the primary analysis had a reported p-value of 0.12, the registry states that secondary tests were presented for illustrative purposes.
12. Kaplan-Meier Estimation and Censoring
Although the posted analyses are summarized primarily through hazard ratios, the underlying endpoints are time-to-event outcomes. Kaplan-Meier estimation is a natural descriptive method for these data because it estimates the probability of remaining event-free over time while accounting for right censoring.
Here, di is the number of events at an observed event time and ni is the number of participants at risk immediately before that time.
The key distinction is that censoring does not mean that a participant experienced the event. Instead, follow-up becomes unavailable for the endpoint after the censoring time. A valid survival analysis therefore uses both event information and the amount of observed follow-up.
The reported median follow-up for the three posted efficacy analyses was 39.9 months. The registry does not report median event-free survival estimates in the information summarized here, so the hazard ratios remain the primary posted effect measures for these analyses.
13. Understanding the Composite Primary Endpoint
The primary endpoint combines four clinically distinct events:
| Component | Role in primary endpoint |
|---|---|
| Cardiovascular death | Qualifying component of the composite |
| Resuscitated cardiac arrest | Qualifying component of the composite |
| Non-fatal myocardial infarction | Qualifying component of the composite |
| Non-fatal stroke | Qualifying component of the composite |
The endpoint is defined by the first occurrence of any component after randomization. This means the primary hazard ratio summarizes the time to the earliest qualifying component rather than the total number of cardiovascular events accumulated by each participant.
Composite endpoints can be statistically efficient because several types of events contribute to the event count. However, the interpretation remains tied to the exact composite definition. A single HR does not establish that the treatment effect is identical for cardiovascular death, myocardial infarction, stroke, and resuscitated cardiac arrest individually.
14. Serious Adverse Events
The registry reports serious adverse events by treatment arm using affected participants and the number at risk.
| Treatment group | Participants with serious adverse events | Participants at risk |
|---|---|---|
| Placebo | 923 | 3,076 |
| Dalcetrapib | 900 | 3,071 |
These figures are descriptive safety results. They should not be treated as a formal efficacy-style treatment effect estimate because the registry information here does not provide a confidence interval, hypothesis test, or time-to-event model for serious adverse events.
The difference between the denominators is also relevant when interpreting the raw affected counts. A comparison of counts alone does not account for the number of participants at risk in each group.
15. Analysis Population and Randomization
The posted statistical analyses specify all subjects randomized to treatment as the analysis population. This is consistent with an intention-to-treat-oriented efficacy comparison, in which participants remain associated with their randomized treatment assignment for the principal treatment comparison.
Why randomization matters
Randomization establishes the treatment assignment before the outcome period begins. This supports a causal comparison because, in expectation, treatment groups are balanced with respect to measured and unmeasured baseline factors.
Why analysis population matters
The interpretation of an effect estimate depends partly on who is included in the analysis. Here, the registry identifies all randomized participants as the analysis population for the posted efficacy analyses.
16. Blinding and Endpoint Adjudication
The study used quadruple masking. The registry also states that all efficacy endpoints were adjudicated by an independent clinical endpoint committee.
These features are particularly relevant for cardiovascular composite endpoints because several components require clinical definitions and adjudication. Independent endpoint adjudication provides a standardized mechanism for determining whether reported clinical events satisfy the prespecified endpoint definitions.
17. Limitations
- COVID-19 pandemic: the registry identifies the occurrence of the COVID-19 pandemic during study conduct as the main limitation of the study.
- Composite endpoint: the primary analysis combines several clinically distinct cardiovascular outcomes into one time-to-first-event endpoint.
- Time-to-event interpretation: hazard ratios are model-based relative measures and are not direct measures of absolute risk reduction.
- Proportional-hazards assumption: interpretation of a single Cox hazard ratio is most straightforward when the relative hazards are reasonably stable over time.
- Multiplicity: the secondary endpoints were subject to a prespecified testing hierarchy, so their reported p-values require the stated primary-endpoint condition to be interpreted as formal confirmatory tests.
- Secondary endpoints: the registry describes the secondary statistical tests as illustrative because the primary analysis did not meet the specified significance condition.
- Safety comparisons: the posted serious-adverse-event information is descriptive and does not include a corresponding formal effect estimate or confidence interval in the reported analysis.
18. Why This Trial Matters Statistically
dal-GenE is a useful teaching example because it combines randomized treatment assignment, masking, clinical endpoint adjudication, composite cardiovascular outcomes, time-to-event analysis, stratified Cox modeling, and multiplicity control within a single phase 3 trial.
| Concept | How it appears in dal-GenE |
|---|---|
| Randomization | Participants were randomized to dalcetrapib or placebo. |
| Blinding | The trial used quadruple masking. |
| Time-to-event endpoint | The primary endpoint measured time from randomization to the first occurrence of a qualifying composite event. |
| Kaplan-Meier framework | Time-to-event outcomes can be described through survival-function estimation and censoring. |
| Cox proportional-hazards model | The primary and secondary posted analyses used stratified Cox models. |
| Hazard ratio | The primary treatment effect was reported as HR 0.88. |
| Confidence interval | The primary 95% CI was 0.75–1.03. |
| P-value | The primary two-sided p-value was 0.12. |
| Stratified analysis | Region and ACS index event type were used as stratification factors in the primary model. |
| Multiplicity adjustment | Hochberg's step-up procedure was specified for formal testing of secondary endpoints conditional on primary success. |
19. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary stratified Cox model estimated an HR of 0.88 with a 95% CI of 0.75–1.03 and a two-sided p-value of 0.12. The interval includes the null value of 1.
Clinical interpretation
The statistical result describes the estimated relative hazard for the defined composite endpoint. Clinical interpretation requires keeping the composite definition, follow-up period, uncertainty interval, and study-conduct limitation in view rather than reducing the result to a single number.
The distinction is important. A statistical estimate is not itself a treatment recommendation. The HR, confidence interval, and p-value describe the evidence generated by the randomized comparison under the specified analysis framework.
20. Reading the Confidence Interval Correctly
The primary HR was 0.88 with a 95% CI of 0.75–1.03. The point estimate lies below 1, but the confidence interval extends through the null value of 1. The interval therefore communicates uncertainty that cannot be captured by the point estimate alone.
A 95% confidence interval should not be described as saying there is a 95% probability that the true hazard ratio lies between 0.75 and 1.03. Rather, it is an interval produced by a procedure that, under repeated sampling and the model assumptions, has the stated coverage property.
The confidence interval communicates both direction and precision, while the p-value summarizes evidence against the specified null hypothesis. Neither one is a direct measure of clinical importance.
21. P-values and the Null Hypothesis
The primary hypothesis was explicitly stated as a two-sided superiority test:
The reported p-value was 0.12. Under the specified model and null hypothesis, this represents the strength of evidence against a hazard ratio equal to 1. It does not establish the probability that the null hypothesis is true.
Likewise, a p-value should not be used as a surrogate for effect size. Two studies can have similar p-values with very different effect estimates, and a statistically small p-value does not automatically imply a clinically large treatment effect.
22. Multiplicity: Why the Secondary Results Need Context
The registry's secondary-endpoint strategy provides a clear example of hierarchical testing. The secondary endpoints were to be formally tested using Hochberg's step-up procedure only when the primary analysis produced a significant treatment effect at p<0.05.
| Stage | Registry-specified role |
|---|---|
| Primary endpoint | Primary superiority analysis |
| Primary significance condition | Primary treatment effect at p<0.05 |
| Secondary endpoints | Formal Hochberg step-up testing only if the primary condition was met |
| If primary condition was not met | Secondary statistical tests presented solely for illustrative purposes |
Because the primary analysis reported p=0.12, the secondary p-values of 0.95 and 0.27 should be understood within this registered framework. The point estimates and confidence intervals remain useful descriptive summaries, but the registry does not treat these secondary tests as formally confirmatory under the stated conditional procedure.
23. Primary Endpoint vs Secondary Endpoints
| Endpoint role | Event definition | HR | 95% CI | P-value |
|---|---|---|---|---|
| Primary | CV death, resuscitated cardiac arrest, non-fatal MI, or non-fatal stroke | 0.88 | 0.75–1.03 | 0.12 |
| Secondary | Primary composite plus ACS hospitalization with ECG abnormalities or unanticipated coronary revascularization | 1.00 | 0.88–1.13 | 0.95 |
| Secondary | Primary composite plus hospitalization for new or worsening heart failure | 0.92 | 0.79–1.07 | 0.27 |
The three estimates should not be interpreted as repeated measurements of one endpoint. Each composite has a different event definition, so the hazard ratio answers a different statistical question.
24. What the Analysis Can and Cannot Establish
The randomized comparison and stratified survival analysis provide an estimate of the relative hazard for the defined time-to-first-event endpoints. They also provide uncertainty intervals around those estimates.
They do not, by themselves, establish that the treatment effect is identical across every component of the composite, that the treatment has the same effect at every point in time, or that the observed hazard ratio represents an absolute reduction in event probability.
Those distinctions are fundamental to interpreting clinical-trial survival analyses. The statistical model is answering a precisely defined question about event timing, treatment assignment, and the specified endpoint.
25. Related Tutorials
Learn more about the methods used in this trial:
26. Related Calculators
27. Sources
- ClinicalTrials.gov: NCT02525939.
- PubMed: PMID 35856777.
- PubMed: PMID 34173811.
- PubMed: PMID 32087417.
Continue through Clinical Biostats
Explore the statistical methods behind randomized trials, survival analysis, endpoint interpretation, and clinical-trial methodology.
28. Record Summary
dal-GenE provides a focused example of randomized cardiovascular time-to-event analysis. The trial randomized 6,147 participants across two treatment arms and used quadruple masking. Its primary endpoint was a time-to-first-event composite of cardiovascular death, resuscitated cardiac arrest, non-fatal myocardial infarction, and non-fatal stroke. The primary analysis used a stratified Cox proportional-hazards model with region and ACS index event type as stratification factors.
The primary hazard ratio was 0.88, with a 95% confidence interval of 0.75–1.03 and a two-sided p-value of 0.12. The two posted secondary analyses produced HRs of 1.00 and 0.92, with 95% confidence intervals of 0.88–1.13 and 0.79–1.07, respectively. The registry specified a Hochberg step-up procedure for formal testing of secondary endpoints only if the primary analysis was significant at p<0.05.
The statistical story is therefore best understood through the combination of randomization, time-to-event endpoint definitions, stratified Cox modeling, hazard ratios, confidence intervals, p-values, and the prespecified multiplicity framework. The COVID-19 pandemic during study conduct is identified by the registry as the main study limitation.