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
EVOLVE was a randomized, double-blind, parallel-group phase 3 trial comparing cinacalcet with placebo in patients with secondary hyperparathyroidism and chronic kidney disease. The registered primary endpoint was a composite time-to-event outcome encompassing all-cause mortality, myocardial infarction, hospitalization for unstable angina, heart failure, or a peripheral vascular event.
| Feature | EVOLVE |
|---|---|
| Phase | Phase 3 |
| Conditions | Secondary Hyperparathyroidism; Chronic Kidney Disease |
| Design | Randomized, double-blind, parallel |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 3,883 |
| Interventions | Cinacalcet and placebo |
| Primary endpoint | One registered time-to-event endpoint |
| Primary hypothesis | Superiority |
| Trial status | Completed |
| Start | August 22, 2006 |
| Primary completion | April 10, 2012 |
| Lead sponsor | Amgen |
2. Clinical Question
The primary statistical question was whether assignment to cinacalcet, compared with placebo, was associated with a different time to the first confirmed event in a composite of all-cause mortality, myocardial infarction, hospitalization for unstable angina, heart failure, or peripheral vascular event.
Population
Participants with secondary hyperparathyroidism and chronic kidney disease enrolled in the EVOLVE phase 3 trial.
Intervention
Cinacalcet hydrochloride therapy.
Comparator
Placebo.
Primary question
Does cinacalcet change the time to the first confirmed event in the registered primary composite endpoint?
3. Trial Design
Cinacalcet
- Cinacalcet hydrochloride therapy
- Randomized treatment assignment
- Included in the ITT efficacy analysis
- Compared with placebo using time-to-event methods
Placebo
- Placebo
- Randomized treatment assignment
- Included in the ITT efficacy analysis
- Reference group for hazard-ratio estimates
4. Endpoints
| Endpoint | Registry definition / time frame | Statistical approach |
|---|---|---|
| Primary composite endpoint | Time to Primary Composite Endpoint (All-cause Mortality, Myocardial Infarction, Hospitalization for Unstable Angina, Heart Failure or Peripheral Vascular Event), from date of randomization until date of first confirmed primary composite endpoint event, assessed up to 5.4 years. | Stratified Cox proportional-hazards model and stratified log-rank test. |
| All-cause mortality | From date of randomization until date of confirmed all-cause mortality endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Myocardial infarction | From date of randomization until date of first confirmed myocardial infarction endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Hospitalization for unstable angina | From date of randomization until date of first confirmed hospitalization for unstable angina endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Heart failure | From date of randomization until date of first confirmed heart failure endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Peripheral vascular event | From date of randomization until date of first confirmed peripheral vascular endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Cardiovascular mortality | From date of randomization until date of first confirmed cardiovascular mortality endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Stroke | From date of randomization until date of first confirmed stroke endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Bone fracture | From date of randomization until date of first confirmed bone fracture endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
| Parathyroidectomy | From date of randomization until date of first confirmed parathyroidectomy endpoint event, assessed up to 5.4 years. | Cox proportional-hazards model and log-rank test. |
The registry reports one primary endpoint and nine secondary time-to-event outcomes among the posted statistical analyses. All of the listed efficacy analyses use the Efficacy Analysis Set, which includes all randomized participants, with participants analyzed according to their randomized treatment group using the intention-to-treat method.
5. Statistical Methodology
Time-to-event framework
The EVOLVE efficacy analyses are fundamentally survival analyses. Rather than asking only whether an event occurred, the endpoint records the time from randomization to the first confirmed event, with follow-up extending up to 5.4 years for the registered outcomes.
This framework is important because participants can have different lengths of observable follow-up. A participant who has not experienced the event by the end of available observation contributes information up to the point of censoring rather than being treated as if the event had occurred.
Intention-to-treat analysis
The registry states that the Efficacy Analysis Set includes all randomized participants and that participants were analyzed in the treatment group as randomized using the ITT method. This preserves the randomized comparison as the foundation for the efficacy analysis.
The key statistical advantage is preservation of the treatment comparison created by randomization. ITT does not mean that every participant necessarily has an observed event; time-to-event methods account for censoring and incomplete event observation.
Stratification
The registry specifies that the primary and secondary analyses were stratified by history of diabetes and country. Stratification allows the treatment comparison to account for these prespecified factors when estimating and comparing time-to-event outcomes.
| Component | Registry specification |
|---|---|
| Stratification factors | History of diabetes; country |
| Analysis population | Efficacy Analysis Set; all randomized participants |
| Assignment for analysis | Treatment group as randomized |
| Primary effect measure | Hazard ratio |
| Primary tests | Cox proportional-hazards model; log-rank test |
| Hypothesis type | Superiority |
Cox proportional-hazards model
The primary endpoint's effect estimate was obtained using a Cox proportional-hazards model. The reported hazard ratio compares cinacalcet with placebo, with values below 1 indicating a lower estimated instantaneous event rate for cinacalcet relative to placebo under the fitted model.
For example, an HR of 0.93 corresponds to an estimated hazard approximately 7% lower for cinacalcet relative to placebo because 1 − 0.93 = 0.07. This is a relative hazard interpretation, not an absolute reduction in the probability of an event.
Log-rank testing
The registry also reports stratified log-rank tests. The log-rank test provides a hypothesis test comparing the time-to-event experience between the randomized groups over follow-up, whereas the Cox model provides a hazard-ratio estimate and its confidence interval.
For the reported log-rank analyses, the registry specifies two-sided testing. The p-value therefore describes the compatibility of the observed treatment-group separation with the null hypothesis under the stated test; it does not measure the magnitude or clinical importance of an effect.
6. Primary Endpoint Result
The primary endpoint was the time from randomization to the first confirmed primary composite endpoint event, consisting of all-cause mortality, myocardial infarction, hospitalization for unstable angina, heart failure, or peripheral vascular event, assessed up to 5.4 years.
Cinacalcet vs placebo: primary composite endpoint
95% CI: 0.85–1.02 · Two-sided log-rank P = 0.112
Analysis: Efficacy Analysis Set; ITT; Cox proportional-hazards model and stratified log-rank test.
| Primary endpoint component | Specification |
|---|---|
| Endpoint | Time to Primary Composite Endpoint: all-cause mortality, myocardial infarction, hospitalization for unstable angina, heart failure or peripheral vascular event |
| Follow-up frame | From date of randomization until date of first confirmed primary composite endpoint event, assessed up to 5.4 years |
| Comparison | Cinacalcet vs placebo |
| Analysis population | Efficacy Analysis Set; all randomized participants |
| Assignment principle | Intent-to-treat |
| Model | Cox proportional-hazards model |
| Stratification | History of diabetes and country |
| Hazard ratio | 0.93 |
| 95% confidence interval | 0.85–1.02 |
| Log-rank P-value | 0.112 |
| Hypothesis | Superiority |
The hazard ratio of 0.93 means that the fitted Cox model estimated the instantaneous rate of experiencing the first primary composite event to be approximately 7% lower in the cinacalcet group than in the placebo group, after accounting for the specified stratification factors.
The estimate alone does not mean that 7% fewer participants experienced an event, nor does it mean that each participant had a 7% lower probability of experiencing the composite endpoint. A hazard ratio is a relative time-to-event measure.
The two-sided 95% confidence interval of 0.85–1.02 expresses uncertainty around the estimated hazard ratio. Because the interval includes 1, the registry's reported estimate is compatible with both a modest reduction and a modest increase in the estimated hazard relative to placebo under the model.
The log-rank P = 0.112 is a hypothesis-test result, not a measure of effect size. It does not tell us that the treatment effect is 11.2%, and a p-value should not be used as a substitute for the hazard ratio or its confidence interval.
The primary endpoint is also composite. Its result represents the time to the first qualifying component event, so the composite does not show that cinacalcet had the same effect on every individual component.
7. Secondary Endpoint Results
The registry reports nine secondary time-to-event outcomes. For each Cox analysis, the hazard ratio is cinacalcet versus placebo and the analysis uses the Efficacy Analysis Set with ITT treatment assignment and stratification by history of diabetes and country.
| Secondary endpoint | HR | 95% CI | Log-rank P |
|---|---|---|---|
| Time to All-cause Mortality | 0.94 | 0.85–1.04 | 0.249 |
| Time to Myocardial Infarction | 0.97 | 0.79–1.19 | 0.800 |
| Time to Hospitalization for Unstable Angina | 0.82 | 0.58–1.18 | 0.283 |
| Time to Heart Failure | 0.82 | 0.68–0.99 | 0.034 |
| Time to Peripheral Vascular Event | 0.87 | 0.72–1.07 | 0.190 |
| Time to Cardiovascular Mortality | 0.92 | 0.80–1.07 | 0.277 |
| Time to Stroke | 1.07 | 0.82–1.40 | 0.607 |
| Time to Bone Fracture | 0.89 | 0.75–1.07 | 0.218 |
| Time to Parathyroidectomy | 0.44 | 0.36–0.54 | <0.001 |
All-cause mortality
The Cox estimate for time to all-cause mortality was HR 0.94, with a 95% CI of 0.85–1.04. The corresponding two-sided log-rank P-value was 0.249. The hazard-ratio estimate is close to 1, while the confidence interval spans both sides of 1.
Myocardial infarction
The estimated hazard ratio for time to myocardial infarction was 0.97 with a 95% CI of 0.79–1.19. The two-sided log-rank P-value was 0.800. The relatively broad interval illustrates the uncertainty associated with an endpoint that is more narrowly defined than the primary composite.
Hospitalization for unstable angina
For time to hospitalization for unstable angina, the reported HR was 0.82 with a 95% CI of 0.58–1.18, and the two-sided log-rank P-value was 0.283. The point estimate is below 1, but the confidence interval includes 1.
Heart failure
The reported HR for time to heart failure was 0.82, with a 95% CI of 0.68–0.99 and a two-sided log-rank P-value of 0.034. The estimate corresponds to an approximately 18% lower estimated hazard under the Cox model because 1 − 0.82 = 0.18.
The heart-failure result is a useful example of why effect estimates, confidence intervals, and p-values should be read together. The HR of 0.82 describes the estimated relative hazard, while the 95% CI of 0.68–0.99 describes uncertainty around that estimate. The P-value of 0.034 describes the reported two-sided hypothesis test.
The p-value does not establish the magnitude of the effect. Nor does the hazard ratio establish the absolute number of heart-failure events prevented. The interpretation also remains subject to the assumptions and censoring structure of the time-to-event analysis.
Peripheral vascular event
The HR for time to peripheral vascular event was 0.87, with a 95% CI of 0.72–1.07. The two-sided log-rank P-value was 0.190. The estimate is below 1, but the confidence interval includes 1.
Cardiovascular mortality
The reported HR for time to cardiovascular mortality was 0.92, with a 95% CI of 0.80–1.07 and a two-sided log-rank P-value of 0.277.
Stroke
The HR for time to stroke was 1.07, with a 95% CI of 0.82–1.40. The two-sided log-rank P-value was 0.607. Unlike most of the reported secondary point estimates, the point estimate is above 1; the confidence interval is also wide enough to include values below and above 1.
Bone fracture
The reported HR for time to bone fracture was 0.89, with a 95% CI of 0.75–1.07 and a two-sided log-rank P-value of 0.218.
Parathyroidectomy
The reported HR for time to parathyroidectomy was 0.44, with a 95% CI of 0.36–0.54. The two-sided log-rank P-value was <0.001.
An HR of 0.44 corresponds to an estimated instantaneous event rate approximately 56% lower for cinacalcet than placebo under the fitted Cox model because 1 − 0.44 = 0.56. The 95% CI of 0.36–0.54 indicates considerably more precision around this estimate than is seen for several of the other secondary endpoints.
Importantly, this does not mean that 56% of participants avoided parathyroidectomy, nor does it provide an absolute risk reduction. The analysis is a time-to-event comparison and must be interpreted in the context of censoring, follow-up, and the treatment assignment framework.
8. Primary Composite Endpoint: Why the Composite Matters
The EVOLVE primary endpoint combines five clinically distinct event categories: all-cause mortality, myocardial infarction, hospitalization for unstable angina, heart failure, and peripheral vascular event. The endpoint is defined as time to the first confirmed qualifying event.
One patient, one first event
A participant contributes the time to the first qualifying component of the composite rather than separate primary events for each component.
Different components
The component outcomes can have different frequencies, clinical implications, and treatment effects. A composite HR therefore does not imply identical effects across all components.
Time-to-event structure
The endpoint incorporates when the first event occurs, rather than simply classifying participants as event or no event at a fixed date.
Interpretation
The primary HR describes the composite endpoint as defined by the registry; it should not be relabeled as an effect on mortality alone.
9. Understanding the Primary Hazard Ratio
The primary HR of 0.93 is a relative measure comparing the instantaneous rate of the first primary composite event between cinacalcet and placebo. Expressed descriptively, the point estimate is consistent with an approximately 7% lower estimated hazard for cinacalcet because 1 − 0.93 = 0.07.
That statement is about the estimated hazard under the Cox model. It is not equivalent to saying that the probability of experiencing the composite endpoint was reduced by 7%, because hazards, risks, and absolute risk differences are different quantities.
The 95% CI of 0.85–1.02 gives the uncertainty range around the reported HR under the analysis framework. It contains 1, so the interval includes a value corresponding to no relative hazard difference between treatment groups.
The width of the interval also matters. A confidence interval should be read not merely as a binary threshold but as an indication of how precisely the treatment effect has been estimated.
The primary two-sided log-rank P-value of 0.112 describes the evidence against the null hypothesis under the specified test. It does not quantify how clinically important the treatment difference is and should not replace the HR and confidence interval.
The primary HR does not establish an absolute risk reduction, a number needed to treat, a probability of individual benefit, or a separate treatment effect for each component of the composite. Those quantities require different information and analyses.
10. Analysis Population and ITT Interpretation
The registry states that the primary and secondary efficacy analyses used the Efficacy Analysis Set, which includes all randomized participants. Participants were analyzed in the treatment group as randomized using the ITT method.
| Question | EVOLVE specification |
|---|---|
| Who enters the efficacy analysis? | All randomized participants in the Efficacy Analysis Set. |
| How are participants assigned for analysis? | According to their randomized treatment group. |
| Primary comparison | Cinacalcet vs placebo. |
| Time-to-event model | Cox proportional-hazards model. |
| Hypothesis test | Stratified log-rank test. |
| Stratification | History of diabetes and country. |
ITT is particularly important in randomized trials because it maintains the comparison created at randomization. It also means that the treatment-effect estimate is an estimate of the effect of assignment to the randomized strategy rather than simply the effect among participants who remained continuously exposed to treatment.
11. Safety: Serious Adverse Events
The registry reports serious adverse events by treatment arm as affected participants over participants at risk.
| Safety measure | Placebo | Cinacalcet |
|---|---|---|
| Serious adverse events, affected / at risk | 1351 / 1923 | 1338 / 1938 |
These figures are presented as reported by the registry. They are not a time-to-event treatment-effect estimate and therefore should not be interpreted in the same way as the primary hazard ratio.
12. Statistical Methods Explained
Why was a Cox proportional-hazards model used?
The registered endpoints are time-to-event outcomes, with follow-up extending up to 5.4 years. A Cox model is designed to compare event hazards while incorporating the timing of events and accommodating right-censored observations. EVOLVE used the model to estimate hazard ratios comparing cinacalcet with placebo.
What does an HR of 0.93 mean?
An HR of 0.93 means that the estimated instantaneous event rate for the cinacalcet group was 0.93 times that of the placebo group under the fitted model. Descriptively, this is an estimated 7% lower hazard. It does not mean a 7% lower probability of experiencing the endpoint or a 7% absolute reduction in risk.
Why was a log-rank test also used?
The log-rank test provides a formal comparison of the time-to-event experience between treatment groups. In EVOLVE, the registry reports a stratified log-rank test alongside the Cox model. The log-rank P-value supplies hypothesis-test evidence, while the Cox model supplies an interpretable relative-effect estimate with a confidence interval.
Why stratify by history of diabetes and country?
The registry specifies stratification by history of diabetes and country. Stratified analysis allows the treatment comparison to account for these factors when comparing the survival experience between randomized treatment groups rather than treating every participant as belonging to a single unstratified risk set.
Why is ITT important for this trial?
ITT analysis keeps participants associated with the treatment group to which they were randomized. This preserves the principal benefit of randomization for efficacy comparisons and avoids redefining the treatment groups based on later treatment behavior.
How should a confidence interval be read?
A 95% confidence interval describes statistical uncertainty around the estimated hazard ratio under the model and sampling framework. For the primary endpoint, the interval of 0.85–1.02 includes 1, while the parathyroidectomy interval of 0.36–0.54 lies entirely below 1. These intervals describe precision around estimates; they do not describe the range of effects experienced by individual participants.
What does the composite endpoint change about interpretation?
The primary endpoint records the time to the first event from a defined set of five component outcomes. Consequently, the HR of 0.93 describes the composite as a whole. It should not be interpreted as though it were a mortality-only hazard ratio or as proof that the treatment had the same effect on each component.
13. Interpreting the Secondary Endpoint Pattern
The secondary results illustrate why a table of hazard ratios should be read as a collection of distinct analyses rather than as one overall statistic. The estimates range from 0.44 for time to parathyroidectomy to 1.07 for time to stroke, with different confidence-interval widths and p-values.
| Endpoint | What the reported estimate describes | 95% CI includes 1? |
|---|---|---|
| All-cause mortality | Relative hazard of confirmed all-cause mortality | Yes |
| Myocardial infarction | Relative hazard of first confirmed myocardial infarction | Yes |
| Hospitalization for unstable angina | Relative hazard of first confirmed hospitalization | Yes |
| Heart failure | Relative hazard of first confirmed heart-failure event | No |
| Peripheral vascular event | Relative hazard of first confirmed peripheral vascular event | Yes |
| Cardiovascular mortality | Relative hazard of first confirmed cardiovascular mortality event | Yes |
| Stroke | Relative hazard of first confirmed stroke event | Yes |
| Bone fracture | Relative hazard of first confirmed bone fracture event | Yes |
| Parathyroidectomy | Relative hazard of first confirmed parathyroidectomy event | No |
The presence or absence of 1 within a confidence interval is only one part of the statistical interpretation. The size of the estimate, its precision, the endpoint definition, the analysis population, and the time-to-event structure all matter. The registry does not provide a multiplicity adjustment specification in the information summarized here, so the secondary p-values should not be treated as though a particular familywise error-control strategy had been documented.
14. Proportional-Hazards Interpretation
The Cox proportional-hazards model is built around a proportional-hazards interpretation: the relative hazard represented by the model is assumed to be adequately summarized by a hazard ratio over the analyzed follow-up.
The hazard ratio is related to the regression coefficient through HR = exp(β). In a two-group comparison, the estimated HR summarizes the relative hazard between the treatment groups under the fitted model.
A hazard ratio is therefore not a fixed risk ratio at a particular calendar time. If hazards vary in a strongly non-proportional manner, a single HR can become a less complete description of the treatment-group separation. The registry reports the Cox proportional-hazards model, but the ClinicalTrials.gov record does not report a formal assessment of the proportional-hazards assumption.
15. Censoring and Follow-up
The registered endpoint definitions specify follow-up from the date of randomization until the date of the first confirmed event, with assessment up to 5.4 years. Participants who do not experience the endpoint during their observable follow-up contribute information until censoring.
Event
The participant experiences the endpoint event being analyzed and contributes the corresponding event time.
Censoring
The participant's event time is not observed during the available follow-up, so the analysis uses the information available up to the censoring point.
Why timing matters
A time-to-event analysis distinguishes an event occurring earlier from an event occurring later rather than reducing both observations to a simple yes/no outcome.
Model dependence
The Cox hazard ratio summarizes the treatment comparison through a model rather than directly reporting an absolute probability difference.
16. Primary Endpoint vs Individual Components
The primary composite includes five components. Several of those components are also registered separately as secondary time-to-event endpoints. This creates an important statistical distinction between the overall composite and its individual components.
| Primary composite component | Also analyzed separately? |
|---|---|
| All-cause mortality | Yes |
| Myocardial infarction | Yes |
| Hospitalization for unstable angina | Yes |
| Heart failure | Yes |
| Peripheral vascular event | Yes |
The composite endpoint can have greater statistical event information than a narrowly defined component because several event types contribute to the endpoint. But its clinical interpretation remains dependent on the relationship among its components. An overall composite result cannot by itself establish that every component moved in the same direction or to the same magnitude.
17. What the P-values Do — and Do Not — Mean
The primary log-rank P-value was 0.112. This is the reported two-sided hypothesis-test result for the comparison of the primary composite time-to-event endpoint. It is not the probability that the null hypothesis is true, and it is not the probability that the observed HR occurred by chance.
The secondary log-rank p-values range from <0.001 for time to parathyroidectomy to 0.800 for time to myocardial infarction. These p-values quantify evidence under their respective statistical tests; they do not provide a common scale for comparing the clinical importance of the endpoints.
A small p-value can accompany a relatively modest effect estimate, while a larger p-value can occur with an estimate that is potentially clinically meaningful but imprecisely estimated. The HR and confidence interval therefore remain central to interpretation.
18. Important Limitations and Interpretation Issues
- Composite endpoint: the primary endpoint combines five different cardiovascular and mortality outcomes, so the primary HR should be interpreted as an effect on the composite rather than on any one component.
- Hazard-ratio interpretation: a Cox HR is a model-based relative measure and should not be converted directly into an absolute risk difference without additional information.
- Proportional-hazards assumption: the registry reports a Cox proportional-hazards model, but the ClinicalTrials.gov record does not report a formal assessment of whether proportional hazards held throughout follow-up.
- Censoring: time-to-event methods depend on the handling and assumptions associated with censoring and incomplete follow-up.
- Secondary endpoints: multiple secondary outcomes were analyzed. The registry information summarized here does not specify a multiplicity-adjustment strategy for these comparisons.
- Subgroup structure: the reported analyses were stratified by history of diabetes and country, but no subgroup-specific treatment-effect estimates are reported in the ClinicalTrials.gov record.
- Absolute effects: the registry information summarized here provides hazard ratios and confidence intervals for the posted Cox analyses but does not provide the absolute event probabilities needed to calculate an absolute risk reduction.
- Endpoint-specific precision: confidence intervals vary substantially across secondary outcomes, reflecting differing statistical information and uncertainty for the individual endpoints.
19. Why This Trial Matters Statistically
EVOLVE is a useful teaching example because the registry results bring together several core principles of clinical-trial survival analysis within a single randomized comparison.
| Concept | How it appears in EVOLVE |
|---|---|
| Randomization | Participants were randomized to cinacalcet or placebo. |
| Blinding | The trial was double-blind. |
| Parallel design | The registry identifies the design model as parallel. |
| ITT analysis | All randomized participants in the Efficacy Analysis Set were analyzed according to randomized treatment assignment. |
| Time-to-event endpoint | The primary outcome measured time to the first confirmed composite event. |
| Hazard ratio | Cox models estimated relative hazards for cinacalcet versus placebo. |
| Confidence interval | 95% two-sided confidence intervals were reported for Cox estimates. |
| Log-rank test | Two-sided log-rank p-values were reported for the primary and secondary time-to-event comparisons. |
| Stratified analysis | Analyses were stratified by history of diabetes and country. |
| Composite endpoint | The primary endpoint combined mortality and several cardiovascular event categories. |
| Secondary endpoints | Individual clinical events were separately evaluated using the same general survival-analysis framework. |
20. Summary of the Statistical Evidence
The primary EVOLVE analysis estimated a hazard ratio of 0.93 for cinacalcet versus placebo for time to the first confirmed primary composite endpoint, with a 95% CI of 0.85–1.02 and a two-sided log-rank P-value of 0.112. The analysis used all randomized participants in the Efficacy Analysis Set, analyzed according to randomized treatment assignment, with stratification by history of diabetes and country.
Among the secondary endpoints, the reported hazard ratios were 0.94 for all-cause mortality, 0.97 for myocardial infarction, 0.82 for hospitalization for unstable angina, 0.82 for heart failure, 0.87 for peripheral vascular event, 0.92 for cardiovascular mortality, 1.07 for stroke, 0.89 for bone fracture, and 0.44 for parathyroidectomy.
The corresponding two-sided log-rank p-values were 0.249, 0.800, 0.283, 0.034, 0.190, 0.277, 0.607, 0.218, and <0.001, respectively. These results illustrate why statistical interpretation should combine the estimated effect, its uncertainty, endpoint definition, and analysis framework rather than relying on p-values alone.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Calculators
23. Sources
- ClinicalTrials.gov: NCT00345839 — EVOLVE.
- PubMed: PMID 29525393.
- PubMed: PMID 27080930.
- PubMed: PMID 26059012.
- PubMed: PMID 25887067.
- PubMed: PMID 25851955.
Continue through the Clinical Biostats statistical library
Explore deeper tutorials and statistical calculators for survival analysis, treatment-effect estimation, clinical-trial design, and statistical inference.
24. Record Summary
EVOLVE provides a clear example of a randomized clinical trial centered on time-to-event methodology. The primary endpoint was a composite of all-cause mortality, myocardial infarction, hospitalization for unstable angina, heart failure, and peripheral vascular event, analyzed using a stratified Cox proportional-hazards model and stratified log-rank test. The primary HR of 0.93 had a 95% CI of 0.85–1.02, while the corresponding two-sided log-rank P-value was 0.112.
The secondary analyses demonstrate the same statistical framework across distinct clinical outcomes. The reported HRs ranged from 0.44 for time to parathyroidectomy to 1.07 for time to stroke. Reading these estimates appropriately requires attention to endpoint definition, censoring, the Cox model, confidence intervals, hypothesis testing, ITT analysis, and the prespecified stratification by history of diabetes and country.