← Clinical Trials
Coronary Artery Disease Phase 3 Time-to-Event Analysis NCT02446990

SIGNIFY: Complete Statistical Analysis of Ivabradine in Coronary Artery Disease

An independent statistical analysis of the randomized phase 3 SIGNIFY trial evaluating ivabradine versus placebo in patients with stable coronary artery disease without clinical heart failure, with emphasis on the prespecified primary composite endpoint and Cox proportional-hazards analysis.

SIGNIFY  ·  Phase 3  ·  19,102 participants  ·  Completed
Scope of this record

This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Trial-specific numerical results on this page are restricted to the data reported for SIGNIFY in the ClinicalTrials.gov record.

1. Trial at a Glance

SIGNIFY was a randomized, quadruple-masked, parallel-group phase 3 trial evaluating ivabradine against placebo in patients with coronary artery disease. The registry reports 19,102 participants and a primary time-to-event composite endpoint defined by the first occurrence of cardiovascular death or non-fatal myocardial infarction.

19,102
Enrollment
2 randomized arms
1.08
Primary HR
95% CI 0.96–1.2
0.1969
Primary P-value
Two-sided
3379 / 9539
Ivabradine serious AEs
Affected / at risk
FeatureSIGNIFY
Trial nameSIGNIFY
NCT identifierNCT02446990
PhasePhase 3
StatusCOMPLETED
ConditionCoronary Artery Disease
AllocationRANDOMIZED
Design modelPARALLEL
MaskingQUADRUPLE
Primary purposeTREATMENT
Enrollment19,102
InterventionsIvabradine; Placebo
Start2009-09
Primary completion2014-01
Lead sponsorInstitut de Recherches Internationales Servier

2. Clinical Question

The primary statistical question was whether the time to the first occurrence of the primary composite endpoint differed between participants assigned to ivabradine and those assigned to placebo.

Population

Patients with coronary artery disease, specifically described in the brief title as having stable coronary artery disease without clinical heart failure.

Intervention

Ivabradine.

Comparator

Placebo.

Primary question

Under a superiority framework, does assignment to ivabradine alter the hazard of the first primary composite event relative to placebo?

3. Trial Design

01
Randomize19,102 participants
02
IvabradineActive intervention
03
PlaceboComparator
04
FollowTime-to-event outcomes
05
AnalyzeCox proportional hazards
Allocation
The registry describes the trial as RANDOMIZED.
Design model
The registry describes a PARALLEL design.
Masking
The registry reports QUADRUPLE masking.
Primary purpose
The registry identifies the primary purpose as TREATMENT.
ARM 1

Ivabradine

  • Intervention type: drug
  • Compared with placebo
  • Primary comparison based on time to the first primary composite event
ARM 2

Placebo

  • Intervention type: drug
  • Comparator for ivabradine
  • Primary comparison based on time to the first primary composite event

4. Primary Endpoint

EndpointRegistry definition / time frameAnalysis
Primary Composite Endpoint The events are expressed as the time to occurrence of the first event, defined as the duration between the date of rando…

Definition: First event among cardiovascular death or non-fatal myocardial infarction
Cox proportional-hazards model; hazard ratio; superiority

The primary endpoint is a time-to-event composite. Rather than simply asking whether a participant eventually experienced one of the component events, the registered analysis considers the time until the first qualifying event. The composite contains two clinically distinct event types: cardiovascular death and non-fatal myocardial infarction.

Why the first event matters. A participant can contribute to the composite endpoint through either component, but the endpoint is defined around the first occurrence. This makes the analysis naturally compatible with survival-analysis methods, because both whether an event occurs and when it occurs contribute information.

5. Statistical Methodology

Cox proportional-hazards model

The registry reports Regression, Cox as the analysis method for the primary endpoint and normalizes this to a Cox proportional-hazards model. The effect measure is a hazard ratio, and the hypothesis type is superiority.

Model concept
h(t) = h0(t) × exp(βX)

For a two-group comparison, the exponentiated treatment coefficient corresponds to a hazard ratio. A hazard ratio below 1 represents a lower estimated instantaneous event rate for the numerator group; a value above 1 represents a higher estimated instantaneous event rate.

Hazard ratio

The hazard ratio compares the instantaneous event rates between the randomized groups over follow-up, conditional on remaining event-free up to the relevant time. It is a relative time-to-event measure, not a probability and not a direct measure of absolute risk.

Confidence interval

The primary analysis reports a two-sided 95% confidence interval. The interval provides a range of values compatible with the estimated treatment effect under the statistical model and sampling framework. It is useful for judging both the direction and precision of the estimated hazard ratio.

Statistical significance

The primary analysis reports a P-value of 0.1969 under a superiority hypothesis. A P-value measures the compatibility of the observed result with the specified null hypothesis; it does not measure the magnitude or clinical importance of the treatment effect.

What the model uses

Time-to-event information, with the event defined as the first occurrence of the registered primary composite endpoint.

What the effect measure is

A hazard ratio from the Cox proportional-hazards model comparing ivabradine with placebo.

What the hypothesis is

Superiority: the analysis tests whether the treatment groups differ in the time-to-event outcome.

What is not reported here

The ClinicalTrials.gov record does not provide a stratification scheme, covariate list, analysis population definition, missing-data procedure, imputation method, or interim-analysis specification.

6. Primary Result

Primary Composite Endpoint

Ivabradine vs placebo

HR 1.08

95% CI: 0.96–1.2   ·   P = 0.1969   ·   Two-sided

Analysis: Cox proportional-hazards model   ·   Hypothesis: superiority

EndpointIvabradine vs placebo95% CIP-valueMethod
Primary Composite Endpoint HR 1.08 0.96–1.2 0.1969 Cox proportional-hazards model
Clinical Biostats interpretation

An estimated hazard ratio of 1.08 means that the fitted Cox model estimated a higher instantaneous rate of the first primary composite event in the ivabradine group relative to placebo, by a factor of 1.08. Expressed differently, the point estimate is 8% above 1 on the hazard-ratio scale.

This does not mean that ivabradine caused 8% more participants to experience the endpoint, nor does it mean that an individual participant had exactly an 8% higher probability of an event. A hazard ratio is a model-based relative time-to-event measure.

The 95% CI of 0.96–1.2 shows that the point estimate has uncertainty that extends on both sides of 1.00. The interval therefore includes values corresponding to a lower hazard as well as values corresponding to a higher hazard for ivabradine relative to placebo.

The P-value of 0.1969 is not an effect-size measure. It describes the statistical evidence against the null hypothesis under the specified superiority framework. It should not be interpreted as a percentage probability that the treatment works or fails to work.

As with any Cox analysis, interpretation of a single hazard ratio depends on the proportional-hazards model being an appropriate description of the relative hazards over time. The ClinicalTrials.gov record does not report a formal assessment of that assumption.

7. Secondary Endpoint Results

The registry contains 14 posted statistical analyses in total: one primary analysis and 13 secondary analyses. All of the secondary analyses reported here use Cox proportional-hazards regression, hazard ratios, two-sided 95% confidence intervals, and superiority hypotheses.

Secondary endpointHR95% CIP-value
All-cause Mortality1.060.94–1.210.3461
Cardiovascular Mortality1.100.94–1.280.2493
Coronary Mortality1.060.89–1.260.5162
Fatal Myocardial Infarction1.350.88–2.050.1647
Non-fatal Myocardial Infarction1.040.90–1.210.6024
Elective Coronary Revascularisation0.890.75–1.040.1458
Coronary Revascularisation (Elective or Not)1.000.89–1.120.9790
Secondary Composite Endpoint 11.060.92–1.220.4299
Secondary Composite Endpoint 20.970.88–1.080.5916
Secondary Composite Endpoint 30.980.89–1.080.6963
Secondary Composite Endpoint 41.060.96–1.180.2222
Secondary Composite Endpoint 51.050.94–1.180.3671
Secondary Composite Endpoint 60.970.87–1.070.5285

The registry supplies the same label, Secondary Composite Endpoint, for the final six secondary analyses rather than providing distinct endpoint names in the ClinicalTrials.gov record. They are therefore numbered in their posted order here rather than assigned names that are not present in the trial data.

Secondary endpoint time frames

EndpointRegistry time frame
All-cause MortalityFrom the date of randomisation to death, up to 48 months
Cardiovascular MortalityFrom the date of randomisation to death, up to 48 months
Coronary MortalityFrom the date of randomisation to death, up to 48 months
Fatal Myocardial InfarctionFrom the date of randomisation to death, up to 48 months
Non-fatal Myocardial InfarctionFrom the date of randomisation to the date of first occurrence of the event, up to 48 months
Elective Coronary RevascularisationFrom the date of randomisation to the date of first occurrence of the event, up to 48 months
Coronary Revascularisation (Elective or Not)From the date of randomisation to the date of first occurrence of the event, up to 48 months
Secondary Composite Endpoint 1–6From the date of randomisation to the date of first occurrence of the event, up to 48 months

8. How to Read the Secondary Hazard Ratios

The secondary results span both sides of a hazard ratio of 1.00. For example, the estimate for elective coronary revascularisation was 0.89, whereas the estimate for fatal myocardial infarction was 1.35. These are point estimates on the same relative time-to-event scale, but their confidence intervals indicate substantial uncertainty for some endpoints.

Several confidence intervals include 1.00. For example, all-cause mortality has a 95% CI of 0.94–1.21, cardiovascular mortality has a CI of 0.94–1.28, and non-fatal myocardial infarction has a CI of 0.90–1.21. A confidence interval crossing 1.00 means that the data are compatible with both a lower and higher hazard under the model.

Do not rank the secondary endpoints by P-value. The registry reports multiple secondary analyses, but the ClinicalTrials.gov record does not specify a multiplicity-adjustment procedure. Consequently, each P-value should be read as the result of its individual reported analysis rather than treating the collection as a simple set of independent confirmatory tests.

9. Safety: Serious Adverse Events

The registry reports serious adverse events by randomized arm using affected participants over participants at risk.

ArmAffectedAt riskRegistry measure
Ivabradine337995393379/9539
Placebo326395443263/9544

The ClinicalTrials.gov record supports reporting these counts, but they do not provide a statistical analysis comparing serious adverse-event rates between arms. The numbers therefore should not be converted here into a risk ratio, risk difference, or P-value.

Safety and efficacy answer different statistical questions. The primary endpoint is a time-to-event efficacy endpoint analyzed with a Cox model. The serious-adverse-event information posted on ClinicalTrials.gov for SIGNIFY is a count of affected participants relative to those at risk. Without a reported comparative analysis, the registry data do not establish a formal statistical difference in serious adverse events between the randomized groups.

10. Statistical Methods Explained

Why was a Cox proportional-hazards model used?

The primary endpoint is explicitly a time-to-event endpoint: the time from randomisation until the first cardiovascular death or non-fatal myocardial infarction. Cox regression is designed to compare time-to-event outcomes while retaining information about when events occur rather than reducing the outcome to a simple yes/no indicator.

What does a hazard ratio of 1.08 mean?

The hazard ratio of 1.08 is the estimated relative hazard for ivabradine versus placebo from the fitted Cox model. A value of 1.08 is above the null value of 1.00, so the point estimate corresponds to a higher estimated instantaneous event rate in the ivabradine group. It does not represent an 8% absolute increase in the proportion of participants experiencing the endpoint.

Why is the confidence interval important?

The 95% confidence interval of 0.96–1.2 communicates precision around the estimated hazard ratio. The interval spans 1.00, so the reported estimate is compatible with a lower hazard as well as a higher hazard for ivabradine under the model. The width of the interval is also important: the interval is more informative than the point estimate alone because it shows how much uncertainty surrounds that estimate.

What does the P-value of 0.1969 tell us?

The P-value quantifies the statistical evidence against the relevant null hypothesis under the specified superiority analysis. It does not tell us the probability that the treatment is effective, and it does not tell us how clinically important an effect is. Effect magnitude and uncertainty are conveyed by the hazard ratio and confidence interval.

How does Kaplan-Meier estimation relate to this type of endpoint?

Kaplan-Meier estimation is a standard descriptive approach for time-to-event data. It estimates the probability of remaining event-free over time while accounting for right censoring. The SIGNIFY learning pathway identifies Kaplan-Meier estimation as a relevant concept, while the posted formal statistical-analysis field identifies Cox regression as the analysis method. The ClinicalTrials.gov record does not provide Kaplan-Meier estimates or a reported Kaplan-Meier curve.

What does a two-sided confidence interval imply?

The posted analyses use two-sided 95% confidence intervals. A two-sided interval allows uncertainty in both directions around the estimated hazard ratio. For a hazard ratio, the null value is 1.00 rather than 0, because 1.00 represents equal hazards between the comparison groups.

Why should the composite endpoint be interpreted as a single outcome?

The primary composite endpoint is defined by the first event among cardiovascular death or non-fatal myocardial infarction. The primary hazard ratio therefore describes the treatment comparison for the composite endpoint as defined, not a pooled average of two separately analyzed effects. The components can have different clinical meanings and event frequencies, so the composite should not automatically be interpreted as though every component behaved identically.

11. Understanding Time-to-Event Analysis

Time-to-event analysis is particularly useful when participants can be followed for different lengths of time or when some participants have not experienced the endpoint by the end of observation. Instead of discarding the timing information, survival-analysis methods incorporate the duration of follow-up.

Conceptual survival function
S(t) = P(T > t)

The survival function represents the probability that the event time T exceeds a particular time t. For a composite endpoint such as the SIGNIFY primary endpoint, the event is the first qualifying component event.

The Cox model works on the hazard scale. The model does not require specification of the baseline hazard function, which is one reason it is widely used for randomized time-to-event comparisons. Its interpretation nevertheless depends on the model being an appropriate representation of the hazard relationship over time.

Event

The first cardiovascular death or non-fatal myocardial infarction for the primary composite endpoint.

Time origin

The registered primary endpoint describes the duration beginning at randomisation.

Effect scale

Hazard ratio comparing ivabradine with placebo.

Statistical model

Cox proportional-hazards regression.

12. Randomization and Causal Interpretation

Randomization is a central design feature of SIGNIFY. The registry identifies the allocation as RANDOMIZED, with two parallel intervention groups. Randomization is important because, when implemented appropriately, treatment assignment is determined independently of participants' prognostic characteristics in expectation.

This creates a foundation for comparing outcomes between the assigned groups. The resulting treatment contrast is different from a purely observational association because the exposure is assigned by the trial design rather than selected solely by patients or investigators.

Randomization does not eliminate every source of uncertainty. Even randomized groups can differ by chance in baseline characteristics, and statistical uncertainty remains in the estimated treatment effect. Randomization establishes the design framework; the confidence interval and analysis model quantify uncertainty around the observed comparison.

13. What the Primary Result Does — and Does Not — Establish

What the estimate says

The fitted Cox model produced a hazard ratio of 1.08 for ivabradine versus placebo for the primary composite endpoint.

What the estimate does not say

It does not say that 8% more participants experienced the endpoint, nor does it describe an individual's probability of experiencing an event.

What the CI says

The 95% CI of 0.96–1.2 expresses uncertainty around the hazard-ratio estimate and crosses the null value of 1.00.

What the P-value says

The P-value of 0.1969 describes statistical evidence under the superiority hypothesis; it is not a measure of effect size.

The most important statistical distinction is between effect magnitude, precision, and evidence against a null hypothesis. SIGNIFY reports all three through the hazard ratio, confidence interval, and P-value, respectively. Reading only one of these quantities would give an incomplete statistical picture.

14. Multiplicity and Multiple Endpoints

The ClinicalTrials.gov record contains one primary endpoint analysis and 13 secondary endpoint analyses. Multiple statistical analyses create a broader inferential context because the probability of observing at least one apparently unusual result can change as the number of analyses increases.

The registry-reported SIGNIFY data identify the hypothesis type for the analyses as superiority, but they do not specify a multiplicity-adjustment procedure, hierarchical testing sequence, alpha-allocation strategy, or other formal error-control mechanism.

Interpretation caution: The individual secondary P-values should not automatically be treated as though each were an independently prespecified confirmatory test at an overall 5% type-I-error level. The ClinicalTrials.gov record does not provide enough information to reconstruct a multiplicity-control strategy.

15. Analysis Features Not Specified in the Supplied Registry Data

The statistical-analysis fields provide a clear description of the primary model and effect measure, but several methodological details are not reported in the ClinicalTrials.gov record.

Design / analysis topicWhat the ClinicalTrials.gov record supports
StratificationNo stratification variables are reported in the ClinicalTrials.gov record.
Analysis populationThe ClinicalTrials.gov record does not define an intention-to-treat, per-protocol, or other formal analysis population.
Missing data / imputationNo missing-data or imputation method is reported in the ClinicalTrials.gov record.
Interim analysisNo interim-analysis procedure is reported in the ClinicalTrials.gov record.
MultiplicityNo multiplicity-adjustment method is reported in the ClinicalTrials.gov record.
Bayesian methodsNo Bayesian method is reported; the posted method is Cox regression.
Non-inferiority marginNot applicable to the reported superiority hypothesis; no non-inferiority margin is reported.
CrossoverNo crossover procedure is reported in the ClinicalTrials.gov record.

These omissions matter because they define how far a reader can go when reconstructing the complete statistical analysis plan from the registry alone. The reported Cox model and hazard ratios are clear; the additional design details above cannot be inferred reliably from the ClinicalTrials.gov record.

16. Secondary Results in Statistical Context

The secondary results provide several examples of why a point estimate should be read together with its confidence interval. The Elective Coronary Revascularisation analysis has an estimated HR of 0.89 with a 95% CI of 0.75–1.04. The point estimate is below 1.00, but the interval extends above 1.00.

For Coronary Revascularisation (Elective or Not), the estimated HR is 1.00 with a 95% CI of 0.89–1.12. Here the point estimate itself is exactly at the hazard-ratio null value, while the interval communicates the uncertainty surrounding that estimate.

For Fatal Myocardial Infarction, the point estimate is 1.35, with a substantially wider 95% CI of 0.88–2.05. The width of this interval illustrates why the point estimate should not be treated as though it were a precise estimate of the underlying treatment effect.

Illustrative analysisPoint estimate95% CIStatistical lesson
Elective Coronary RevascularisationHR 0.890.75–1.04A point estimate below 1.00 can coexist with an interval that crosses 1.00.
Coronary Revascularisation (Elective or Not)HR 1.000.89–1.12The point estimate equals the null value while the interval shows uncertainty in both directions.
Fatal Myocardial InfarctionHR 1.350.88–2.05A larger point estimate can have substantial uncertainty when the interval is wide.

17. Clinical Biostats Statistical Reading Guide

A useful way to read the SIGNIFY results is to proceed in a fixed sequence rather than beginning with the P-value.

01
EndpointWhat event is being analyzed?
02
TimeWhen does the event count?
03
EffectWhat is the hazard ratio?
04
PrecisionWhat is the CI?
05
EvidenceWhat is the P-value?

For the primary endpoint, this sequence produces a coherent statistical story: the endpoint is a first-event composite; the outcome is analyzed as time-to-event; the hazard ratio is 1.08; the two-sided 95% confidence interval is 0.96–1.2; and the reported P-value is 0.1969.

This approach avoids the common mistake of treating a P-value as the primary description of a treatment effect. The hazard ratio describes the estimated relative effect, the confidence interval describes its uncertainty, and the P-value addresses the specified hypothesis test.

18. Why This Trial Matters Statistically

SIGNIFY is a useful teaching case because the registry data bring together the core elements of randomized survival analysis without requiring the reader to rely on a binary endpoint alone.

ConceptHow it appears in SIGNIFY
RandomizationThe trial is registered as RANDOMIZED.
Parallel-group designThe design model is PARALLEL.
BlindingThe registry reports QUADRUPLE masking.
Time-to-event endpointThe primary endpoint is defined by time to the first cardiovascular death or non-fatal myocardial infarction.
Cox regressionThe primary and secondary analyses use Cox regression.
Hazard ratioThe reported effect measure is a hazard ratio.
Confidence intervalEach posted analysis reports a two-sided 95% confidence interval.
P-valueEach posted statistical analysis includes a P-value.
Superiority testingThe primary and secondary analyses identify the hypothesis type as superiority.
Composite endpointThe primary endpoint combines cardiovascular death and non-fatal myocardial infarction through the first-event definition.
Multiple endpointsThe registry contains one primary analysis and 13 secondary analyses.
Safety reportingSerious adverse events are reported as affected participants over participants at risk by arm.

The trial therefore provides a compact example of an important statistical principle: the design, endpoint definition, analysis model, effect measure, and uncertainty measure must be interpreted together. A hazard ratio cannot be understood independently of the event definition, and a P-value cannot substitute for the effect estimate and confidence interval.

19. Important Limitations and Interpretation Issues

20. Record Timeline

2009-09

Trial start

The registry lists the SIGNIFY study start as September 2009.

2014-01

Primary completion

The registry lists January 2014 as the primary completion date.

Completed

Registry status

The trial status is reported as COMPLETED.

21. Statistical Interpretation vs Clinical Interpretation

Statistical interpretation

The primary Cox model produced HR 1.08 with a two-sided 95% CI of 0.96–1.2 and P = 0.1969. The confidence interval includes the hazard-ratio null value of 1.00.

Clinical interpretation

The registry data alone do not provide enough information to translate the primary hazard ratio into absolute event rates, numbers needed to treat, or other absolute clinical-effect measures.

This distinction is important. Statistical evidence concerns the estimated treatment contrast and its uncertainty under a specified model. Clinical interpretation requires understanding the endpoint, absolute event burden, duration of follow-up, patient population, safety profile, and other contextual information. The registry-reported SIGNIFY data do not provide all of those quantities.

22. A Closer Look at the Primary Composite

ComponentRole in primary endpoint
Cardiovascular deathOne of the two events that can constitute the first primary composite event.
Non-fatal myocardial infarctionOne of the two events that can constitute the first primary composite event.
First qualifying eventThe primary endpoint is based on the first occurrence of either component.

The use of a first-event composite makes the endpoint suitable for a time-to-event framework. Importantly, the hazard ratio of 1.08 applies to this composite definition. It should not be relabeled as an effect estimate for cardiovascular death alone or myocardial infarction alone.

23. Related Tutorials

Learn more about the methods used in this trial:

24. Related Statistical Calculators

25. Sources

Continue through Clinical Biostats

Explore the statistical methods behind randomized trials, survival analysis, confidence intervals, hypothesis testing, and clinical-trial design.

26. Record Summary

SIGNIFY provides a clear example of randomized clinical-trial survival analysis. The registry describes a completed phase 3, randomized, parallel-group, quadruple-masked trial with 19,102 participants comparing ivabradine with placebo in patients with stable coronary artery disease without clinical heart failure. Its primary endpoint was a time-to-event composite of the first occurrence of cardiovascular death or non-fatal myocardial infarction, analyzed using a Cox proportional-hazards model.

The reported primary hazard ratio was 1.08, with a 95% CI of 0.96–1.2 and a P-value of 0.1969. The most statistically informative reading of this result combines all three quantities: the hazard-ratio point estimate, the uncertainty expressed by the confidence interval, and the evidence reported by the hypothesis test.

The secondary analyses illustrate the same principle across several time-to-event outcomes. Estimates ranged from below 1 to above 1, while confidence intervals varied in width. Because the ClinicalTrials.gov record does not identify a multiplicity-adjustment procedure or provide the full analysis plan, secondary results should be interpreted in the context of the complete set of analyses rather than by ranking their individual P-values.

Clinical Biostats methodology: A trial-results page should distinguish what the registry actually reports from what statistical principles allow us to explain. For SIGNIFY, the registry provides the trial design, endpoint definitions, Cox regression method, hazard ratios, confidence intervals, P-values, and serious-adverse-event counts. Additional methodological details not present in the ClinicalTrials.gov record is intentionally not reconstructed.