← Clinical Trials
Type 2 Diabetes Mellitus Phase 3 Randomized NCT00032487

VADT: Complete Statistical Analysis of Glycemic Control in Type 2 Diabetes

An independent statistical analysis of the randomized phase 3 VADT trial evaluating intensive versus standard glycemic control in people with type 2 diabetes, with primary analysis of time to the first major macrovascular event.

Trial period: December 1, 2000–May 30, 2008  ·  Enrollment: 1791  ·  Status: 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.

1. Trial at a Glance

VADT was a phase 3, randomized, parallel-group trial in type 2 diabetes mellitus. The ClinicalTrials.gov record describes a comparison of Arm 1 versus Arm 2, with the primary analysis focused on time from baseline to the first major macrovascular event up to 82 months.

1791
Enrolled
Phase 3 trial
2
Arms
Parallel design
0.79
Hazard ratio
95% CI .79–.99
<0.05
P-value
Log-rank analysis
FeatureVADT
Trial nameVADT
Brief titleGlycemic Control and Complications in Diabetes Mellitus Type 2 (VADT)
PhasePhase 3
ConditionType 2 Diabetes Mellitus
AllocationRandomized
DesignParallel
MaskingNone
Primary purposeTreatment
Enrollment1791
Arms2
StatusCompleted
Lead sponsorVA Office of Research and Development
Sponsor typeFED
ClinicalTrials.govNCT00032487

2. Clinical Question

The ClinicalTrials.gov record identifies the clinical comparison as intensive glycemic control versus standard glycemic control and specify a primary cardiovascular composite outcome. The primary statistical question was whether the two randomized groups could be considered equivalent with respect to the hazard of experiencing a first major macrovascular event.

Population

Participants with type 2 diabetes mellitus enrolled in the VADT phase 3 trial.

Intervention framework

The statistical analysis notes identify an intensive glycemic control group compared with a standard control group.

Comparator

The primary analysis compares Arm 1 with Arm 2. The ClinicalTrials.gov record does not specify a separate arm-level assignment of the listed medications.

Primary question

What is the relative hazard of a first major macrovascular event between the two randomized groups, and does the prespecified equivalence analysis support the intended comparison?

3. Trial Design

01
Randomize 1791 participants
02
Two arms Arm 1 vs Arm 2
03
Follow Post-baseline events
04
Assess Major macrovascular events
05
Analyze Log-rank and hazard ratio
Allocation
Randomized allocation to two parallel arms.
Masking
None.
Primary purpose
Treatment.
Follow-up window
The primary endpoint was assessed from post-baseline time to the first major macrovascular event up to 82 months.
ARM 1

Randomized comparison group

The registry-reported statistical analysis compares Arm 1 with Arm 2. The registry data list insulin, glimepiride, rosiglitazone, and metformin among the interventions, but do not provide enough arm-level detail to assign individual drugs to this arm.

ARM 2

Randomized comparison group

The registry-reported statistical analysis compares Arm 2 with Arm 1. The analysis notes characterize the comparison in terms of intensive versus standard glycemic control.

4. Endpoints

EndpointRegistry definitionTime frameStatistical role
Primary Major Macrovascular Events Myocardial infarction (MI), intervention for coronary artery or Peripheral Vascular Disease (PVD), severe inoperable Coronary Artery Disease (CAD), new or worsening Congestive Heart Failure (CHF), stroke, Cardiovascular (CV) death, or amputation for ischemic gangrene. Post baseline time to the first major macrovascular event up to 82 months Primary endpoint

The endpoint is a time-to-event composite. The event is defined as the first qualifying major macrovascular event after baseline. Because several clinically different outcomes contribute to the composite, the statistical result summarizes the time to the first occurrence of any qualifying component rather than providing a separate treatment effect for each component.

5. Statistical Methodology

Log-rank test

The registry reports a log-rank test for the primary time-to-event comparison. The log-rank test compares the survival experience of two groups across the observed follow-up period while accounting for the timing of events and right-censoring.

Primary comparison
Arm 1  vs  Arm 2  →  time to first major macrovascular event

The reported statistical method is a log-rank test, with a Cox proportional-hazard effect measure reported alongside it.

Cox proportional-hazard effect measure

The reported effect measure is a Cox proportional hazard, normalized here as a hazard ratio. The estimate is 0.79, with a two-sided 95% confidence interval of .79 to .99.

Hazard-ratio interpretation
HR = 0.79  →  estimated hazard ratio for Arm 1 relative to Arm 2

An HR of 0.79 corresponds to an estimated 21% lower instantaneous event hazard for Arm 1 relative to Arm 2, because 1 − 0.79 = 0.21. This is a relative time-to-event measure, not a statement that 21% of participants avoided an event.

Equivalence analysis

The ClinicalTrials.gov record identifies the hypothesis type as non-inferiority or equivalence and state specifically that the statistical analysis tested for equivalence. The analysis notes state that the study assumed 86% power with 21% of effect size and a sample size of 1700 with a 5% drop out rate.

No numerical equivalence margin is reported in the ClinicalTrials.gov record. Therefore, this page does not reconstruct or invent an equivalence margin. In an equivalence design, the confidence interval is normally evaluated against prespecified upper and lower margins; the ClinicalTrials.gov record does not provide those numerical boundaries.

6. Primary Result: Major Macrovascular Events

The primary endpoint was time from baseline to the first major macrovascular event up to 82 months. The registry reports a log-rank analysis comparing Arm 1 with Arm 2 and a Cox proportional-hazard effect measure.

Primary hazard ratio

0.79

95% CI: .79–.99   ·   Two-sided CI   ·   P <0.05

Endpoint: time to the first major macrovascular event up to 82 months

Primary endpointArm 1Arm 2Reported analysis
Primary Major Macrovascular Events Reference comparison Reference comparison HR 0.79; 95% CI .79–.99; P <0.05
Clinical Biostats interpretation

The reported hazard ratio of 0.79 means that the estimated instantaneous rate of a first major macrovascular event in Arm 1 was 0.79 times the corresponding rate in Arm 2 under the reported Cox model. Expressed as a relative reduction, 1 − 0.79 = 0.21, or a 21% lower estimated hazard.

The hazard ratio does not mean that 21% of participants were protected from an event, that the probability of an event was reduced by exactly 21%, or that every participant experienced the same reduction. It is a model-based relative measure of event hazard over time.

The 95% confidence interval of .79–.99 describes statistical uncertainty around the estimated hazard ratio. Its interpretation is about the underlying treatment-effect parameter under the specified statistical framework; it is not a range containing 95% of individual patient effects.

The P <0.05 value addresses evidence against the relevant null hypothesis under the reported test framework. A p-value does not measure the size, clinical importance, or certainty of the observed effect. The magnitude of the estimated association is described by the hazard ratio and its confidence interval.

For an equivalence analysis, however, a conventional p-value should not be treated as a substitute for the prespecified equivalence margins. The ClinicalTrials.gov record states that equivalence was tested but do not provide the numerical margins. Accordingly, the equivalence conclusion cannot be reconstructed beyond the information reported here.

Finally, a Cox hazard ratio relies on the proportional-hazards framework. The ClinicalTrials.gov record identifies the Cox proportional-hazard effect measure but do not provide a diagnostic assessment of that assumption.

7. Understanding the Primary Composite Endpoint

The primary endpoint combines several major macrovascular outcomes: myocardial infarction, coronary or peripheral vascular intervention, severe inoperable coronary artery disease, new or worsening congestive heart failure, stroke, cardiovascular death, and amputation for ischemic gangrene.

Composite componentRegistry definition
Myocardial infarctionMI
Vascular interventionIntervention for coronary artery or Peripheral Vascular Disease (PVD)
Severe coronary diseaseSevere inoperable Coronary Artery Disease (CAD)
Heart failureNew or worsening Congestive Heart Failure (CHF)
StrokeStroke
Cardiovascular deathCardiovascular (CV) death
AmputationAmputation for ischemic gangrene

For a composite endpoint, the reported hazard ratio applies to the time until the first qualifying event. It therefore does not establish that every component had the same effect. A treatment comparison could produce a composite result through effects distributed differently across its component outcomes.

8. Statistical Methods Explained

Why was a log-rank test used?

The primary endpoint is a time-to-event outcome: participants are followed until the first qualifying macrovascular event or until their observation is censored. The log-rank test is designed for comparing time-to-event distributions between randomized groups while using the available follow-up information.

What does a hazard ratio of 0.79 mean?

An HR of 0.79 means the estimated hazard in Arm 1 was 0.79 times the hazard in Arm 2 under the reported Cox model. The simple relative interpretation is a 21% lower estimated hazard. It is not equivalent to a 21-percentage-point reduction in event probability.

What does the 95% confidence interval of .79–.99 tell us?

The confidence interval gives a range of values describing uncertainty around the estimated hazard ratio. The interval is relatively close to 1, so the precision of the estimate matters when interpreting the magnitude. A confidence interval should be considered together with the point estimate rather than replaced by it.

Why is the p-value not the same thing as effect size?

The reported P <0.05 indicates evidence against the relevant null hypothesis under the specified testing framework. It does not tell us how large the treatment effect is. The HR of 0.79 provides the effect estimate, while the confidence interval provides information about its statistical precision.

Why does equivalence require a margin?

Equivalence testing asks whether the observed difference is sufficiently small to fall within a prespecified range considered compatible with equivalence. That range is determined before analyzing the data. The registry-reported VADT registry data state that equivalence was tested but do not provide the numerical equivalence margin, so the margin-based decision rule cannot be reproduced from the ClinicalTrials.gov record.

Why does the composite endpoint matter statistically?

The endpoint combines multiple types of macrovascular events into a single time-to-first-event outcome. This can increase the number of observed events and provide a unified primary outcome, but the resulting hazard ratio describes the composite rather than each individual component. The ClinicalTrials.gov record does not provide component-specific statistical analyses.

What does randomization contribute?

Randomization assigns participants to the comparison groups according to the trial's allocation process rather than allowing treatment assignment to be determined by participant or investigator choice. This is the key design feature supporting a causal comparison between the randomized groups, subject to the usual limitations of trial conduct, follow-up, and generalizability.

9. Confidence Intervals and the Primary Estimate

Reported hazard-ratio interval
Lower CI
.79
Point estimate
0.79
Upper CI
.99

The interval extends from .79 to .99. The point estimate sits at the lower boundary reported for the confidence interval. Because the ClinicalTrials.gov record gives the interval exactly as .79–.99, those values are retained without rounding or recalculation.

Important distinction: the confidence interval is not a prediction interval for future participants. It quantifies uncertainty around the estimated population-level hazard ratio under the statistical model and sampling framework used for the analysis.

10. Equivalence Testing in This Trial

The registry classifies the hypothesis type as non-inferiority or equivalence, while the statistical-analysis comment specifically states: “The statistical analysis tested was for equivalence.”

What the registry reports

The analysis tested equivalence and assumed 86% power with 21% of effect size and a sample size of 1700 with 5% drop out rate.

What is not reported

No numerical equivalence margin is included in the ClinicalTrials.gov record. The margin therefore cannot be reconstructed from this record.

Why the margin matters

Equivalence is a margin-based hypothesis. The confidence interval must be evaluated relative to the prespecified acceptable range, not merely relative to a conventional significance threshold.

Interpretive caution

The reported HR and P-value can be described exactly, but the ClinicalTrials.gov record is insufficient to independently reproduce the complete equivalence decision rule.

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

Relative effect

The reported HR of 0.79 represents the estimated ratio of the event hazards for Arm 1 versus Arm 2. In relative terms, 0.79 corresponds to a 21% lower estimated hazard for Arm 1.

Not an absolute risk difference

The HR does not say that the absolute probability of experiencing a major macrovascular event fell by 21 percentage points. No absolute event probabilities or event counts by randomized arm are reported in the ClinicalTrials.gov record.

Not an individual treatment effect

The HR is a population-level statistical estimate. It does not imply that each individual participant experienced a 21% reduction in their personal probability or hazard.

Precision

The 95% CI of .79–.99 provides the uncertainty interval reported for the estimate. The interval is important because the point estimate alone does not communicate the statistical precision of the treatment comparison.

12. Safety Results

The ClinicalTrials.gov record reports serious adverse events by randomized arm as affected participants divided by participants at risk.

Safety measureArm 1Arm 2
Serious adverse events144/899197/892
Serious adverse events by arm
Arm 1
144/899
Arm 2
197/892

The registry reports these serious-adverse-event counts and denominators but does not provide a formal statistical comparison for this safety measure in the ClinicalTrials.gov record. Therefore, the page does not attach a p-value, hazard ratio, or confidence interval to the serious-adverse-event comparison.

Safety interpretation: serious adverse events are summarized separately from the primary efficacy endpoint. The available safety data do not establish whether any observed numerical difference between the arms is statistically significant or clinically attributable to the randomized intervention.

13. Limitations

The registry itself identifies two important limitations. First, the study population consisted of veterans and was predominantly male, so extrapolation of the findings to women must be undertaken with caution. Second, changes in therapeutic agents occurred after the protocol was designed.

Population generalizability

Because participants were veterans and predominantly men, the registry specifically cautions that extrapolation to women should be done with caution.

Changes in therapy

The registry notes that changes in therapeutic agents occurred since the design of the protocol. This creates an important context for interpreting an older trial framework in relation to later therapeutic practice.

Composite endpoint

The primary outcome combines multiple macrovascular events. The registry-reported statistical analysis reports the composite but does not provide separate component-specific estimates.

Equivalence margin unavailable

The ClinicalTrials.gov record states that equivalence was tested but does not give a numerical equivalence margin, limiting independent reconstruction of the full equivalence decision.

14. Statistical Reporting Limitations

15. Why This Trial Matters Statistically

VADT provides a useful teaching example because its primary analysis combines randomization, a composite time-to-event endpoint, log-rank testing, a Cox hazard ratio, confidence-interval interpretation, and an equivalence hypothesis.

ConceptHow it appears in VADT
RandomizationThe trial used randomized allocation to two parallel arms.
Time-to-event endpointThe primary outcome is time from baseline to the first major macrovascular event up to 82 months.
Composite endpointMI, vascular intervention, severe inoperable CAD, CHF, stroke, CV death, or amputation for ischemic gangrene contribute to the primary outcome.
Log-rank testThe registry reports a log-rank comparison for the primary endpoint.
Hazard ratioThe reported Cox proportional-hazard effect measure is 0.79.
Confidence intervalThe reported two-sided 95% CI is .79–.99.
P-valueThe primary analysis reports P <0.05.
Equivalence testingThe registry states that the statistical analysis tested for equivalence and reports an assumed 86% power.
GeneralizabilityThe registry cautions about extrapolation to women because the veteran population was predominantly male.
Changing therapyThe registry notes that therapeutic agents changed since protocol design.

16. Reading the Primary Result Step by Step

Step 1 · Endpoint

Identify what was measured

The endpoint is time from baseline to the first major macrovascular event up to 82 months, using a composite definition.

Step 2 · Comparison

Identify the randomized groups

The registry analysis compares Arm 1 with Arm 2. The analysis notes describe the treatment framework as intensive glycemic control versus standard control.

Step 3 · Test

Read the log-rank result

The reported method is the log-rank test, which compares the observed time-to-event experience between the randomized groups.

Step 4 · Effect estimate

Read the hazard ratio

The Cox proportional-hazard estimate is 0.79. Relative to 1, this corresponds to a 21% lower estimated hazard for Arm 1.

Step 5 · Uncertainty

Read the confidence interval

The two-sided 95% confidence interval is .79–.99. It communicates uncertainty around the estimated hazard ratio.

Step 6 · Hypothesis framework

Check the equivalence design

The registry states that equivalence was tested. Because the numerical equivalence margin is not reported, the complete margin-based decision cannot be independently reproduced from the available record.

17. Statistical Interpretation vs Clinical Interpretation

Statistical interpretation

The reported primary analysis gives an HR of 0.79 with a two-sided 95% CI of .79–.99 and P <0.05 from the reported log-rank framework. The analysis was identified as an equivalence test.

Clinical interpretation

The primary endpoint is a composite of major macrovascular outcomes. The statistical result describes the time-to-first-composite-event comparison; it does not establish separate effects for every component.

These are deliberately different levels of interpretation. The statistical result describes the estimated relative time-to-event comparison and its uncertainty. Clinical interpretation requires consideration of the endpoint definition, population, safety information, generalizability, and the limitations identified by the registry.

18. Results Scope

The ClinicalTrials.gov record identifies 2 posted outcome measures and 1 posted statistical analysis, with the single formal analysis corresponding to the primary endpoint. The ClinicalTrials.gov record therefore supports a detailed primary-endpoint analysis and a descriptive safety summary, but it does not provide enough information to construct additional endpoint-specific statistical result sections.

Data discipline: This page intentionally does not add median survival, subgroup estimates, component-specific event counts, baseline characteristics, interim analyses, missing-data methods, stratification factors, Bayesian methods, crossover information, or other analyses that are not contained in the registry-reported VADT trial data.

19. Trial Timeline

December 1, 2000

Trial start

The VADT trial began on December 1, 2000.

May 30, 2008

Primary completion

The registry lists May 30, 2008 as the primary completion date.

Completed

Registry status

The registry profile lists the trial status as completed.

20. Interventions Listed in the Registry

the ClinicalTrials.gov record lists four drug interventions:

InterventionRegistry classification
InsulinDrug
GlimepirideDrug
RosiglitazoneDrug
MetforminDrug

The ClinicalTrials.gov record does not provide an arm-by-arm mapping of these listed interventions. Accordingly, the statistical analysis on this page uses the registry's Arm 1 versus Arm 2 comparison and the analysis note's intensive-versus-standard glycemic-control terminology rather than assigning individual medications to either arm.

21. Serious Adverse Events: Statistical Perspective

The serious-adverse-event data are reported as affected participants divided by the number at risk: 144/899 in Arm 1 and 197/892 in Arm 2.

What can be concluded from the ClinicalTrials.gov record
Arm 1: 144/899    |    Arm 2: 197/892

These are descriptive arm-level safety counts. Without a reported statistical comparison, they should not be converted into a formal claim of statistical difference.

The safety denominator is also distinct from the overall enrollment of 1791. The registry specifically reports 899 and 892 participants at risk for this safety measure, so those denominators should be retained rather than replacing them with the total enrollment.

22. What Would Normally Be Examined in a Time-to-Event Analysis?

For an endpoint such as time to first major macrovascular event, the statistical analysis naturally has several connected components:

The registry-reported VADT data explicitly support the log-rank test, Cox hazard ratio, confidence interval, and equivalence framework. They do not provide enough numerical information to reconstruct a Kaplan-Meier curve or a complete margin-based equivalence calculation.

23. Why the Equivalence Framework Changes the Interpretation

A conventional superiority analysis and an equivalence analysis ask different questions. A superiority analysis generally asks whether one group differs from another in a specified direction. An equivalence analysis asks whether the observed difference is sufficiently small to remain inside prespecified acceptable bounds.

Superiority-style question

Is there evidence that the randomized groups differ beyond the null value?

Equivalence-style question

Is the uncertainty interval sufficiently contained within prespecified equivalence margins?

VADT registry information

The ClinicalTrials.gov record explicitly identifies the statistical analysis as an equivalence test.

Key missing design value

The numerical equivalence margin is not included in the ClinicalTrials.gov record, so the complete equivalence criterion cannot be reproduced here.

24. Related Tutorials

Learn more about the methods used in this trial:

25. Related Calculators

26. Sources

Continue through Clinical Biostats statistical methods

Explore the survival-analysis, trial-design, and statistical-inference concepts that underpin randomized clinical-trial analysis.

27. Record Summary

VADT is a phase 3 randomized, parallel-group trial in type 2 diabetes mellitus with 1791 participants and two study arms. Its registry-reported primary analysis evaluates time from baseline to the first major macrovascular event up to 82 months, using a log-rank test and a Cox proportional-hazard effect measure. The reported hazard ratio is 0.79, with a two-sided 95% confidence interval of .79–.99 and P <0.05. The registry identifies the statistical hypothesis as an equivalence analysis and states an assumed 86% power with 21% of effect size and a sample size of 1700 with a 5% drop out rate. The numerical equivalence margin is not contained in the ClinicalTrials.gov record, so the full margin-based equivalence calculation cannot be independently reconstructed.

The principal statistical lesson is that the hazard ratio, confidence interval, p-value, and hypothesis framework answer different questions. The HR describes the estimated relative event hazard, the confidence interval describes statistical precision, the p-value describes evidence against the relevant null hypothesis under the reported test, and the equivalence framework requires prespecified margins. Together, these elements provide a more complete interpretation than any one statistic alone.

Clinical Biostats methodology: This page separates reported registry results from statistical interpretation and does not add numerical trial results that are absent from the registry-reported VADT data.