← Clinical Trials
Cardiovascular Phase 4 Randomized Trial NCT01985360

ISCHEMIA-CKD: Complete Statistical Analysis of Invasive vs Conservative Management in Chronic Kidney Disease

An independent statistical review of the randomized ISCHEMIA-CKD trial, focusing on its time-to-event endpoints, hazard-ratio analysis, cumulative event rates, trial design, and interpretation of the reported results.

Trial status: COMPLETED  ·  Start: 2014-01  ·  Primary completion: 2019-06
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. Numerical trial facts on this page are restricted to the ClinicalTrials.gov record.

1. Trial at a Glance

ISCHEMIA-CKD was a randomized, parallel-group, open-label phase 4 trial enrolling 777 participants with cardiovascular and kidney disease. The two randomized strategies were an invasive strategy and a conservative strategy.

777
Enrollment
Randomized trial
2
Arms
Parallel design
1.01
3-Year HR
95% CI 0.79–1.29
3 years
Primary analysis frame
Cumulative event rate
FeatureISCHEMIA-CKD
Trial nameISCHEMIA-CKD
Brief titleISCHEMIA-Chronic Kidney Disease Trial
PhasePhase 4
StatusCOMPLETED
Enrollment777
AllocationRandomized
DesignParallel
MaskingNone
Primary purposeTreatment
Number of arms2
Lead sponsorNYU Langone Health
ClinicalTrials.govNCT01985360

2. Clinical Question

The trial compared an invasive strategy with a conservative strategy in a population with cardiovascular and kidney disease. The registry identifies cardiovascular diseases, coronary artery disease, heart diseases, myocardial ischemia, kidney disease, and end stage renal failure on dialysis among the trial conditions.

Population

Participants enrolled in the ISCHEMIA-CKD trial, with conditions including cardiovascular disease, coronary artery disease, myocardial ischemia, kidney disease, and end stage renal failure on dialysis.

Intervention strategy

The registry identifies the invasive strategy as one of the two randomized groups. Registered interventions included cardiac catheterization, coronary artery bypass graft surgery, percutaneous coronary intervention, lifestyle, and medication.

Comparator strategy

The conservative strategy was the other randomized group. The ClinicalTrials.gov record identifies the strategy but do not provide a detailed arm-by-arm treatment description.

Primary question

How did the invasive strategy compare with the conservative strategy with respect to death from any cause or myocardial infarction over the registered follow-up periods?

3. Trial Design

01
Enroll777 participants
02
Randomize2 parallel strategies
03
FollowTime-to-event outcomes
04
AssessDeath or myocardial infarction
05
CompareHazard ratio
Allocation
Randomized. Randomization is the principal design feature supporting a comparison between the two treatment strategies.
Design model
Parallel. Participants were assigned to one of two parallel strategies rather than being exposed sequentially to both.
Masking
None. The ClinicalTrials.gov record identifies the trial as unmasked.
Primary purpose
Treatment. The registry classifies the primary purpose as treatment.
ARM 1 · 388 AT RISK FOR REPORTED SERIOUS AE COUNT

Invasive Strategy

  • Randomized invasive strategy.
  • Registered interventions include cardiac catheterization.
  • Registered interventions also include coronary artery bypass graft surgery and percutaneous coronary intervention.
  • Lifestyle and medication are also listed among registered interventions.
ARM 2 · 389 AT RISK FOR REPORTED SERIOUS AE COUNT

Conservative Strategy

  • Randomized conservative strategy.
  • The ClinicalTrials.gov record does not provide a more detailed arm-specific treatment sequence.
  • The reported serious-adverse-event denominator is 389 for this strategy.

4. Trial Timeline

2014-01

Trial start

The ClinicalTrials.gov record lists January 2014 as the trial start.

2019-06

Primary completion

The ClinicalTrials.gov record lists June 2019 as the primary completion date.

Completed

Results posted

The trial is marked COMPLETED and the registry contains posted outcome measures and one posted statistical analysis.

5. Endpoints

Registered primary endpointTime frameRegistry information
Incidence of Death From Any Cause or Myocardial Infarction 2.2 years Results posted. No formal statistical analysis for this endpoint is included in the registry-reported statistical-analyses record.
Cumulative Event Rate of Death From Any Cause or Myocardial Infarction 3 years Results posted and a formal statistical analysis is posted.

How the 3-year cumulative event rate is defined

The registry defines this measure as the estimated cumulative probability of experiencing death from any cause or myocardial infarction within the indicated timeframe in each treatment group. The registry states that its interpretation is similar to Kaplan-Meier event rates and that estimates are expressed as percentages from 0% to 100%.

Endpoint distinction: the 2.2-year endpoint is described as the incidence of death from any cause or myocardial infarction, whereas the 3-year endpoint is explicitly described as a cumulative event rate. Both address the same clinical event combination but are registered at different time frames.

6. Statistical Methodology

The ClinicalTrials.gov record identifies the principal effect measure for the posted formal analysis as a hazard ratio. The record does not name the statistical method, so it does not identify the exact statistical procedure used to generate the reported hazard ratio.

Statistical featureWhat the ClinicalTrials.gov record shows
Formal statistical analysis1 statistical analysis posted
Primary-endpoint analyses1
Analysis with estimate + CI1
Effect measureHazard ratio
Hypothesis typeOther / not stated
CI95%, two-sided

What would normally be used for this endpoint?

A time-to-event endpoint such as death from any cause or myocardial infarction is commonly summarized using Kaplan-Meier estimation, with treatment groups compared using a time-to-event hypothesis test and a Cox proportional-hazards model when a hazard ratio is the effect measure. These are standard analytical approaches for this endpoint type, but the ClinicalTrials.gov record does not state that any particular one of these methods generated the posted ISCHEMIA-CKD estimate.

Methodology boundary: the exact statistical method used for the posted ISCHEMIA-CKD hazard ratio should not be inferred from the effect measure alone. The record does not name the method.

7. Results: Three-Year Cumulative Event Rate

The registry reports a formal statistical analysis for the 3-year cumulative event rate of death from any cause or myocardial infarction, comparing the invasive strategy with the conservative strategy.

Hazard ratio for the 3-year endpoint

1.01

95% CI: 0.79–1.29   ·   Two-sided CI

Comparison: Invasive Strategy vs Conservative Strategy

EndpointTime frameComparisonEffect measureEstimate95% CI
Cumulative Event Rate of Death From Any Cause or Myocardial Infarction 3 years Invasive Strategy vs Conservative Strategy Hazard Ratio 1.01 0.79–1.29
Clinical Biostats interpretation

The reported hazard ratio of 1.01 is very close to 1. Under a hazard-ratio interpretation, it represents an estimated hazard in the invasive-strategy group that is approximately 1% higher than the estimated hazard in the conservative-strategy group.

That statement is a description of the relative hazard estimate. It does not mean that 1% more patients experienced the endpoint, that the absolute event probability differed by 1%, or that an individual patient's risk changed by 1%.

The 95% confidence interval of 0.79–1.29 is important because it is substantially wider than the point estimate alone. It spans values below 1 and above 1, meaning the data represented by this interval are compatible with different relative hazard values on either side of the null value.

The confidence interval describes uncertainty around the estimated treatment effect under the statistical framework; it is not a range in which the effect for individual patients is expected to fall.

No p-value is provided in the registry-reported statistical-analyses data. A p-value would address evidence against a specified null hypothesis; it would not measure the magnitude or clinical importance of the hazard ratio. The hazard ratio and its confidence interval therefore provide the principal quantitative information available here.

8. Results: The 2.2-Year Primary Endpoint

The registry lists Incidence of Death From Any Cause or Myocardial Infarction at 2.2 years as a primary endpoint and indicates that results were posted. However, the registry-reported statistical-analyses record does not contain a formal analysis for this endpoint.

What can be concluded from the ClinicalTrials.gov record: the endpoint was registered, its time frame was 2.2 years, and results were posted. The ClinicalTrials.gov record does not provide a formal effect estimate, confidence interval, or p-value for this endpoint, so no additional numerical comparison is reproduced here.

How this type of endpoint is typically analyzed

For a time-to-event endpoint involving death or myocardial infarction, an analysis would commonly account for the time from randomization to the first qualifying event and for censoring among participants who have not experienced the event by the end of observed follow-up. Kaplan-Meier methods can estimate event probabilities over time, while a Cox model can summarize the relative hazard using a hazard ratio.

Those are methodological explanations rather than claims about the method actually used in this registry analysis. The ClinicalTrials.gov record does not identify the formal method for the 2.2-year endpoint.

9. Statistical Methods Explained

What does a hazard ratio of 1.01 mean?

A hazard ratio compares the estimated instantaneous event rates between two groups within a time-to-event model. An HR of 1 represents equal estimated hazards. An HR of 1.01 therefore lies extremely close to that reference value. It does not directly describe cumulative incidence, absolute risk difference, or the probability that a particular patient experiences the endpoint.

Why is the confidence interval more informative than the point estimate alone?

The point estimate is only one estimate from the observed data. The 95% confidence interval of 0.79–1.29 communicates the statistical uncertainty surrounding the reported HR of 1.01. Because the interval extends both below and above 1, the interval does not isolate a single direction of relative hazard.

Why is the cumulative event rate different from the hazard ratio?

A cumulative event rate describes the estimated probability of experiencing the endpoint by a specified time. A hazard ratio is a relative measure comparing instantaneous event rates within a time-to-event model. They summarize different aspects of the same underlying event process and should not be substituted for one another.

Why does censoring matter?

Time-to-event studies frequently contain participants whose complete event history is not observed during the study period. A participant who has not experienced the event by the last observation may contribute follow-up information without being counted as having experienced the event. Statistical methods such as Kaplan-Meier estimation are designed to incorporate this right-censored information.

Why should the exact analysis method not be assumed?

The ClinicalTrials.gov record reports a hazard ratio but does not name the statistical method. An effect measure can suggest the general class of analysis, but it does not establish every modeling detail. For example, it does not by itself establish whether the reported analysis used a particular Cox-model specification, stratification scheme, or other implementation detail.

Does the confidence interval tell us whether the treatment is clinically important?

No. Statistical precision and clinical importance are related but distinct questions. The confidence interval describes uncertainty around the estimated relative effect. Determining clinical importance also requires understanding the absolute event rates, clinical context, outcome severity, follow-up, and the consequences of the intervention. The ClinicalTrials.gov record does not provide enough numerical information to make all of those assessments.

10. Randomization and Comparative Interpretation

Randomization is central to the statistical interpretation of ISCHEMIA-CKD. Because participants were randomized to the invasive or conservative strategy, the treatment-group comparison is designed to reduce systematic differences in baseline characteristics that could otherwise confound an observational comparison.

What randomization supports

It supports a comparison in which treatment assignment, rather than a participant's clinical characteristics or treatment choice, determines the initial strategy assignment.

What randomization does not guarantee

Randomization does not make the observed groups numerically identical in every characteristic, nor does it eliminate statistical uncertainty in the estimated treatment effect.

Why the analysis is comparative

The reported HR is explicitly based on an invasive-strategy versus conservative-strategy comparison.

Why follow-up matters

The registered primary endpoints are defined at specific time frames, making the duration and handling of follow-up part of the statistical definition of the outcome.

11. Understanding the 3-Year Endpoint

The 3-year outcome is particularly useful for statistical teaching because the registry describes it explicitly as a cumulative event rate. The measure estimates the cumulative probability of death from any cause or myocardial infarction within the indicated timeframe in each treatment group.

Conceptual survival-analysis relationship
Cumulative event probability by time t = 1 − estimated event-free survival at time t

The registry states that the cumulative event-rate measure is interpreted similarly to Kaplan-Meier event rates. The ClinicalTrials.gov record does not provide the individual patient event and censoring records needed to reconstruct the underlying survival curve.

This distinction matters because a cumulative event rate is an absolute time-specific quantity, whereas the reported HR of 1.01 is a relative time-to-event quantity. Presenting both types of measures allows a reader to ask two different questions: how often did the endpoint occur by a specified time, and how did the event rate compare between the randomized strategies over follow-up?

No reconstructed survival curve: the ClinicalTrials.gov record does not include individual event times, censoring times, numbers at risk, or group-specific cumulative event-rate values. A Kaplan-Meier curve should therefore not be fabricated from the single reported hazard ratio.

12. Safety Results

The ClinicalTrials.gov record includes serious adverse events by randomized strategy. The registry reports affected participants and the number at risk for each arm.

Safety measureInvasive StrategyConservative Strategy
Serious adverse events75/38861/389

These counts describe serious adverse events and their corresponding reported denominators. They should be kept conceptually separate from the efficacy endpoint of death from any cause or myocardial infarction. A safety count is not automatically equivalent to the primary efficacy event definition.

Safety interpretation

The reported serious-adverse-event counts are 75/388 for the invasive strategy and 61/389 for the conservative strategy. The ClinicalTrials.gov record does not provide a confidence interval, hypothesis test, relative-risk estimate, or hazard ratio for serious adverse events, so no formal comparative safety inference is added here.

The denominators are also not identical to the overall enrollment of 777. This is an important reminder that safety analyses can use an analysis population or risk set that differs from the total number enrolled in the trial.

13. What the Hazard Ratio Does — and Does Not — Mean

Point estimate

An HR of 1.01 means the estimated instantaneous event rate in the invasive-strategy group was approximately 1% higher than the corresponding estimated event rate in the conservative-strategy group under the reported hazard-ratio framework.

It does not mean that the cumulative event rate was 1% higher, that one additional participant out of every 100 experienced the endpoint, or that an individual's risk changed by 1%.

Confidence interval

The 95% CI of 0.79–1.29 describes uncertainty around the estimated HR. Because it includes the null value of 1, the interval does not establish a precise direction of relative hazard based on the reported estimate alone.

P-value

No p-value is included in the registry-reported statistical analysis. A p-value would quantify the compatibility of the observed data with a specified null hypothesis under the relevant testing framework. It would not quantify the size of the treatment effect and would not replace the HR or its confidence interval.

Model assumptions

When a Cox proportional-hazards model is used, interpretation of a single HR is linked to the model's assumptions about the relationship of hazards over time. Because the ClinicalTrials.gov record does not name the method, it also does not establish which specific proportional-hazards diagnostics or model specifications were used.

14. Missing Data, Censoring, and Analysis Details

The ClinicalTrials.gov record does not report a missing-data or imputation strategy, a censoring rule, stratification factors, or a Bayesian analysis. These omissions are important because a statistical analysis page should distinguish between what is known from the registry and what would merely be typical practice.

Design topicInformation in the ClinicalTrials.gov record
Missing-data / imputation methodNot reported
Censoring rulesNot reported
Stratification factorsNot reported
Bayesian methodsNot reported
Interim analysis detailsNot reported
Non-inferiority marginNot reported
Multiplicity procedureNot reported
Crossover detailsNot reported

For a time-to-event endpoint, censoring is particularly important because participants can contribute information without experiencing the endpoint during observed follow-up. However, the exact censoring convention used in ISCHEMIA-CKD should not be inferred from the fact that the endpoint is time-to-event.

15. Non-Inferiority, Superiority, and Hypothesis Interpretation

The registry-reported statistical analysis lists the hypothesis type as Other / not stated. The available data therefore do not establish a superiority hypothesis, a non-inferiority margin, or a formal equivalence framework for the posted analysis.

What the HR can show

The HR summarizes the relative event rate represented by the reported analysis and provides a quantitative estimate around which uncertainty can be described.

What the HR cannot show alone

An HR by itself does not identify the prespecified hypothesis, the clinical margin, or the decision rule used by the investigators.

This distinction is especially important when interpreting an HR close to 1. A reader should not label such a result "non-inferior" or "equivalent" unless the trial's prespecified hypothesis, margin, confidence-interval rule, and analysis population support that conclusion.

16. Multiplicity and Interim Analysis

The ClinicalTrials.gov record does not report an interim-analysis procedure or a multiplicity-adjustment strategy. There are two registered primary endpoints, but the ClinicalTrials.gov record contains only one formal primary-endpoint analysis.

FeatureRegistry-supported interpretation
Number of registered primary endpoints2
Number of posted outcome measures2
Number of posted statistical analyses1
Primary-endpoint analyses posted1
Multiplicity procedureNot reported
Interim-analysis procedureNot reported

Having two registered primary endpoints does not, by itself, reveal how type I error was controlled. If multiple hypotheses are tested, the interpretation of nominal statistical evidence depends on the prespecified testing hierarchy or multiplicity procedure. That information is not included in the ClinicalTrials.gov record.

17. Analysis Populations and Denominators

The overall enrollment is 777, while the reported serious-adverse-event denominators are 388 and 389. This illustrates why analysis populations should be identified explicitly rather than assuming that every endpoint uses the same denominator.

QuantityReported valueStatistical relevance
Total enrollment777Overall trial size in the registry profile
Invasive Strategy serious-AE denominator388Denominator for the reported serious-adverse-event count
Conservative Strategy serious-AE denominator389Denominator for the reported serious-adverse-event count

The ClinicalTrials.gov record does not specify the exact efficacy analysis population used for the HR of 1.01. Consequently, the HR should be interpreted as the posted registry analysis without assigning it an ITT, per-protocol, as-treated, or other population label that is not explicitly provided.

18. Limitations

19. Why This Trial Matters Statistically

ISCHEMIA-CKD is a useful statistical teaching case because it demonstrates how a randomized clinical trial can combine a clinically meaningful composite time-to-event endpoint with a hazard-ratio analysis while leaving important methodological details outside the information available in a registry summary.

ConceptHow it appears in ISCHEMIA-CKD
RandomizationThe allocation is randomized, creating a comparative framework for the two strategies.
Parallel designTwo randomized strategies are compared in parallel.
Time-to-event analysisDeath from any cause or myocardial infarction is registered with specified follow-up time frames.
Kaplan-Meier estimationThe registry explicitly describes the cumulative event-rate measure as similar in interpretation to Kaplan-Meier event rates.
Hazard ratioThe posted formal analysis reports HR 1.01 with a 95% CI of 0.79–1.29.
Confidence intervalThe two-sided 95% CI quantifies uncertainty around the HR estimate.
Composite endpointThe primary event combines death from any cause and myocardial infarction.
Analysis-population awarenessSafety denominators differ from total enrollment, illustrating why denominators must be tied to the specific analysis.
Registry-method limitationsThe exact statistical method, hypothesis type, and several design-analysis details are not reported in the ClinicalTrials.gov record.

20. A Practical Reading of the ISCHEMIA-CKD Result

A disciplined statistical reading starts with the endpoint definition rather than the numerical estimate. The registered 3-year endpoint is the cumulative event rate of death from any cause or myocardial infarction. The posted formal comparison is summarized by an HR of 1.01 with a 95% CI of 0.79–1.29.

The point estimate is close to the null value of 1, but the confidence interval is the better guide to the precision of the estimate. It spans both directions around the null. That means the estimate should not be presented as evidence of a precisely measured increase or decrease in the underlying event hazard without additional statistical information.

The result also illustrates why an HR should not be interpreted as an absolute event probability. The registry describes a separate cumulative event-rate quantity for the 3-year endpoint, but the ClinicalTrials.gov record does not give the two group-specific cumulative event-rate percentages. Therefore, the available result supports a relative hazard interpretation but does not support construction of an absolute risk difference from the registry-reported numbers.

Finally, the absence of a reported p-value should not be filled in by calculation from the HR and confidence interval. Although a p-value can sometimes be approximated from a confidence interval under particular assumptions, doing so would introduce a calculation and testing convention that is not part of the ClinicalTrials.gov record. This page therefore reports the HR and confidence interval exactly as provided.

21. Related Tutorials

Learn more about the methods used in this trial:

22. Related Calculators

23. Sources

Continue with the statistical methods behind this trial

Explore tutorials on survival analysis, hazard ratios, confidence intervals, randomization, and time-to-event endpoints, then apply those concepts with statistical calculators.

24. Record Summary

ISCHEMIA-CKD provides a useful example of how to read a randomized time-to-event trial when the registry supplies a formal hazard-ratio estimate but only limited information about the underlying statistical implementation. The central posted result is an HR of 1.01 for the 3-year cumulative event rate of death from any cause or myocardial infarction, with a two-sided 95% CI of 0.79–1.29. The registry also identifies a 2.2-year primary endpoint for the same clinical event combination, but the ClinicalTrials.gov record does not provide a formal comparison for that endpoint.

The most important statistical lesson is to keep the different layers of evidence separate: randomization defines the comparative design; the endpoint definition and time frame define what was measured; the hazard ratio describes the relative time-to-event comparison; the confidence interval describes uncertainty around that estimate; and the absence of a reported method, p-value, or additional event-rate detail limits how far the numerical interpretation can be taken.

Clinical Biostats methodology: A trial-results page should distinguish reported registry evidence from standard statistical explanation. Where the ClinicalTrials.gov record does not identify an analysis detail, this page does not infer one simply because it would be conventional for the endpoint.