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
SYNTAX was a randomized, parallel-group, phase 3 treatment trial in coronary artery disease. The registry describes two treatment arms: a polymer-based paclitaxel-eluting TAXUS Express2-SR stent and coronary artery bypass surgery. The registered primary endpoints are three binary outcomes assessed at 12 months.
| Feature | SYNTAX |
|---|---|
| Trial name | SYNTAX |
| Brief title | SYNTAX Study: TAXUS Drug-Eluting Stent Versus Coronary Artery Bypass Surgery for the Treatment of Narrowed Arteries |
| Phase | Phase 3 |
| Condition | Coronary Artery Disease |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 1800 |
| Arms | 2 |
| Results posted | Yes |
| Outcome measures posted | 13 |
| Lead sponsor | Boston Scientific Corporation |
| Sponsor type | INDUSTRY |
| ClinicalTrials.gov | NCT00114972 |
2. Clinical Question
The registry identifies coronary artery disease as the condition under study and compares two treatment strategies: a polymer-based paclitaxel-eluting TAXUS Express2-SR stent and coronary artery bypass surgery. The primary statistical questions are represented by three binary outcomes measured 12 months after enrollment.
Population
Participants enrolled in a phase 3 treatment trial for coronary artery disease and narrowed arteries.
Intervention
Polymer-based Paclitaxel-Eluting TAXUS Express2-SR Stent.
Comparator
Coronary Artery Bypass Surgery.
Primary questions
How do the randomized treatment groups compare with respect to 12-month binary MACCE, the 12-month composite safety endpoint, and repeat revascularization?
3. Trial Design
TAXUS drug-eluting stent
- Polymer-based Paclitaxel-Eluting TAXUS Express2-SR Stent
- Device intervention
Coronary artery bypass surgery
- Coronary Artery Bypass Surgery
- Procedure intervention
4. Primary Endpoints
The registry lists three primary endpoints. All three are binary outcomes with a 12-month post-enrollment time frame. The endpoint definitions below preserve the registry wording while making their statistical structure explicit.
| Primary endpoint | Time frame | Registry definition | Type |
|---|---|---|---|
| Primary Clinical Endpoint of 12-Month Binary MACCE | 12 months post enrollment | Number of participants at primary clinical endpoint of 12-Month binary MACCE. MACCE is defined as: all cause death, cerebrovascular event (stroke), cocumented myocardial infarction, repeat revascularization (PCI and/or CABG). | Binary |
| 12-month Composite Safety Endpoint | 12 months after enrollment | Number of participants at 12-month composite safety endpoint. Composite safety endpoint combines: all cause death, cerebrovascular event (stroke), and documented myocardial infarction. | Binary |
| Repeat Revascularization (PCI and/or CABG) | 12 Months post enrollment | Number of participants with repeat revascularization (PCI and/or CABG). | Binary |
Why the composite endpoint deserves attention
MACCE combines several clinically distinct events: all-cause death, cerebrovascular event (stroke), documented myocardial infarction, and repeat revascularization. Statistically, a composite endpoint can increase the number of observed events and therefore provide more information for a treatment comparison. Clinically, however, its components are not necessarily interchangeable. A treatment difference in the composite can be driven disproportionately by one component, particularly if that component is more frequent than the others.
The safety composite is narrower: it combines all-cause death, cerebrovascular event (stroke), and documented myocardial infarction. Repeat revascularization is listed separately as a primary binary endpoint. This means the registry distinguishes the broader clinical composite from the safety composite and from the revascularization component.
5. Statistical Methodology
The ClinicalTrials.gov record identifies the primary endpoints as binary and confirms that results have been posted, but it contains no formal statistical analyses. Consequently, the ClinicalTrials.gov record does not provide a formal treatment-effect estimate, confidence interval, or p-value for any of the three primary endpoints.
For a binary endpoint, common comparative measures include a risk difference, risk ratio, or odds ratio. A formal analysis can also use a chi-square test or Fisher's exact test, depending on the prespecified analysis plan and the observed cell counts. These are descriptions of standard statistical approaches, not claims about the particular analysis performed in SYNTAX.
Why the endpoint type determines the analysis
Each registered primary endpoint is defined as a participant either reaching the endpoint or not reaching it within the stated 12-month assessment framework. That makes the basic statistical object a binary proportion rather than a continuous measurement or a time-to-event estimate.
A simple comparison of two proportions answers a different question from a Kaplan-Meier or Cox analysis. A time-to-event method uses information about when events occur and can accommodate censoring. The ClinicalTrials.gov record identifies these primary endpoints as binary and does not provide a time-to-event statistical analysis specification.
Composite endpoints and component analysis
For MACCE, the primary binary variable is whether a participant experienced any of the specified component events by the endpoint assessment. A formal analysis should distinguish the composite result from the individual component results. Otherwise, a statistically detectable composite difference could be difficult to interpret clinically if the treatment groups differ mainly in repeat revascularization rather than death, stroke, or myocardial infarction.
Multiplicity across three primary endpoints
Three primary endpoints create a multiple-testing question. If each endpoint were tested independently at the same nominal significance level, the probability of at least one false-positive finding across the family could exceed the nominal level for a single test. A prespecified trial analysis would therefore ordinarily need to define how the three primary endpoints are related statistically, for example through a hierarchical testing strategy or another multiplicity-control procedure.
The ClinicalTrials.gov record does not report a multiplicity procedure, endpoint hierarchy, alpha allocation, or adjusted p-values. Those details should not be inferred from the existence of three registered primary endpoints.
6. Statistical Methods Explained
Why is randomization important for a binary endpoint?
Randomization creates the primary basis for comparing the treatment groups. If the groups are randomized, the observed difference in the 12-month proportion experiencing MACCE can be interpreted within a causal treatment-comparison framework, subject to adherence, follow-up, missing-data, and other trial limitations. Randomization does not guarantee identical groups; chance imbalance remains possible.
What does a binary primary endpoint mean?
A binary endpoint classifies each participant according to whether the specified event occurred within the registered endpoint framework. For SYNTAX, the three primary endpoints are all recorded as binary, so the central quantities are the numbers or proportions of participants with the endpoint in each treatment arm.
Why might risk difference and risk ratio tell different stories?
A risk difference expresses the absolute difference between two event proportions. A risk ratio expresses one event proportion relative to the other. For example, the same relative treatment effect can correspond to very different absolute differences depending on the baseline event rate. Reporting both can therefore make the magnitude of a binary treatment effect easier to understand.
Why does a composite endpoint need component-level interpretation?
A composite endpoint treats any qualifying component as an event for the primary analysis. That increases the event count but combines outcomes that may differ in frequency and clinical importance. A useful interpretation therefore asks not only whether the composite differs between groups, but also which components contribute to that difference.
Why does having three primary endpoints create a multiplicity issue?
Each formal hypothesis test carries some chance of a false-positive result. Testing several primary hypotheses without a prespecified error-control strategy can increase the overall familywise false-positive probability. The appropriate interpretation therefore depends on the protocol or statistical analysis plan's method for controlling or structuring those tests.
Why can't the posted results be treated as a formal statistical comparison here?
The ClinicalTrials.gov record says that results are posted and that 13 outcome measures are posted, but no formal statistical analyses were posted.
7. Planned Analysis
Because the ClinicalTrials.gov record contains no formal statistical-analysis entries, this page does not report numerical efficacy results for the three primary endpoints. Instead, this section explains how endpoints of the registered type would ordinarily be analysed.
| Registered endpoint | Typical statistical analysis | What the analysis would estimate |
|---|---|---|
| 12-Month Binary MACCE | Comparison of binary event proportions between randomized treatment groups; common measures include risk difference, risk ratio, or odds ratio. | Relative or absolute difference in the proportion experiencing MACCE by the 12-month endpoint. |
| 12-month Composite Safety Endpoint | Comparison of binary event proportions between randomized treatment groups. | Difference between groups in the proportion experiencing death, cerebrovascular event (stroke), or documented myocardial infarction. |
| Repeat Revascularization (PCI and/or CABG) | Comparison of binary event proportions between randomized treatment groups. | Difference between groups in the proportion experiencing repeat revascularization. |
What would a confidence interval add?
A confidence interval around a risk difference, risk ratio, or odds ratio would describe statistical uncertainty around the estimated treatment contrast. It would not describe the range of individual patient outcomes. The interval's width would also depend on the amount of information available for the particular endpoint.
What would a p-value add?
A p-value would quantify the compatibility of the observed data with a specified null hypothesis under a particular statistical model or test. It would not measure the size or clinical importance of the treatment effect. A small p-value can accompany a modest absolute difference, while a potentially meaningful difference can have substantial uncertainty in a smaller effective sample.
8. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment arm using an affected/at-risk format. These are the only arm-specific safety results provided in the trial data.
| Treatment arm | Serious adverse events | Reported format |
|---|---|---|
| CABG | 391/897 | Affected / at risk |
| PCI With DES | 357/903 | Affected / at risk |
The affected/at-risk format is useful because it preserves both the number of participants with the safety event and the denominator used for that arm. The ClinicalTrials.gov record does not provide a formal statistical comparison, confidence interval, or p-value for serious adverse events.
9. Results Interpretation: What Can and Cannot Be Concluded
SYNTAX has posted outcome measures, but the ClinicalTrials.gov recordset does not include the numerical values for the three primary endpoint results or their formal statistical analyses. The appropriate statistical conclusion from the registry record is therefore about the structure of the evidence, not the direction or magnitude of an efficacy effect.
What is documented
The trial is randomized, parallel, phase 3, has 2 arms, enrolled 1800.0 participants, and registered three binary primary endpoints at 12 months.
What is not reported
No formal primary-endpoint estimates, confidence intervals, p-values, multiplicity procedure, or statistical-analysis entries are contained in the ClinicalTrials.gov record.
What can be analysed
The endpoint definitions, treatment allocation structure, binary analysis framework, composite construction, and reported serious adverse-event counts can be examined statistically.
What should not be inferred
The existence of posted outcomes does not justify reconstructing treatment effects that are not present in the ClinicalTrials.gov record.
10. Causal Interpretation of the Randomized Comparison
Randomization is central to the causal interpretation of SYNTAX. Because the registry identifies randomized allocation, the intended comparison is between treatment strategies assigned through the trial rather than between observationally selected groups.
That advantage does not remove every source of uncertainty. A treatment comparison can still be affected by missing outcome information, deviations from assigned treatment, differential follow-up, or other post-randomization processes. The ClinicalTrials.gov record does not describe a missing-data or imputation strategy, so no particular approach should be attributed to SYNTAX here.
The lack of masking is also part of the registered design. The registry records masking as none. For outcomes involving clinical events and procedures, lack of masking can be relevant to assessment and management processes, although the ClinicalTrials.gov record does not quantify any resulting bias.
This is the basic statistical architecture of the registered primary analysis. The validity of the final inference depends on how the trial collected, defined, and analysed those outcomes.
11. Composite Endpoint Logic
The primary clinical endpoint is 12-month binary MACCE, defined by the registry as all-cause death, cerebrovascular event (stroke), documented myocardial infarction, or repeat revascularization (PCI and/or CABG). Statistically, a participant meeting any qualifying component is classified as having the composite endpoint.
| Composite | Components specified in the registry | Statistical implication |
|---|---|---|
| 12-Month Binary MACCE | All cause death; cerebrovascular event (stroke); documented myocardial infarction; repeat revascularization (PCI and/or CABG) | Any qualifying component makes the binary composite positive. |
| 12-month Composite Safety Endpoint | All cause death; cerebrovascular event (stroke); documented myocardial infarction | Any qualifying safety component makes the binary composite positive. |
The two composites overlap but are not identical. MACCE includes repeat revascularization, whereas the registered composite safety endpoint does not. This distinction matters because an analysis of the two endpoints can produce different treatment contrasts even when death, stroke, and myocardial infarction are unchanged.
12. Endpoint Hierarchy and Multiplicity
Three outcomes are explicitly labelled primary endpoints. From a statistical-design perspective, that raises two separate questions: whether each endpoint is considered independently important and how the overall type I error is controlled when multiple hypotheses are tested.
| Issue | Why it matters | Information in the ClinicalTrials.gov record |
|---|---|---|
| Three primary endpoints | Multiple confirmatory hypotheses can increase the familywise false-positive probability if each is tested without adjustment. | Three primary endpoints are registered. |
| Multiplicity procedure | A prespecified method can control or structure the overall inferential error rate. | Not reported in the ClinicalTrials.gov record. |
| Endpoint hierarchy | A hierarchical strategy can determine which findings are formally confirmatory and in what order. | Not reported in the ClinicalTrials.gov record. |
| Adjusted p-values | Adjustment can reflect the multiple-hypothesis structure of the analysis. | No p-values are reported. |
13. Analysis Populations and Missing Data
The ClinicalTrials.gov record does not define an intention-to-treat population, per-protocol population, as-treated population, or a separate safety-analysis population beyond the arm-specific serious-adverse-event denominators reported in the ClinicalTrials.gov record.
Similarly, the ClinicalTrials.gov record does not describe how missing 12-month endpoint information was handled. For binary outcomes, missingness can matter because the estimated event proportion depends on how participants without an ascertainable endpoint are handled.
Intention-to-treat
A conventional randomized-trial principle is to analyse participants according to assigned treatment. The ClinicalTrials.gov record does not explicitly state that this was the analysis population for SYNTAX.
Per-protocol
A per-protocol analysis can address outcomes among participants meeting specified treatment and follow-up criteria, but no such population is defined in the ClinicalTrials.gov record.
Missing outcomes
The ClinicalTrials.gov record does not report the extent or handling of missing 12-month endpoint observations.
Imputation
No imputation method is reported in the ClinicalTrials.gov record.
These omissions are important because a binary endpoint is particularly sensitive to the classification of participants whose endpoint status cannot be established. Complete-case, worst-case, best-case, or model-based approaches can produce different estimates. None should be attributed to SYNTAX without supporting trial documentation.
14. Serious Adverse Events and Denominators
The safety data supplies both affected and at-risk counts. The denominators are different between the two arms: 897 for CABG and 903 for PCI With DES. That denominator information should be retained when interpreting the reported serious-adverse-event counts.
These values are reported exactly as reported in the registry. No percentage conversion or formal between-group statistical test is presented because the ClinicalTrials.gov record does not contain one.
For a formal safety comparison, investigators would ordinarily define the safety population, endpoint coding rules, observation period, and statistical method in advance. The meaning of the denominator is especially important: an affected/at-risk count is only interpretable when the population represented by the denominator is clearly defined.
15. Trial Timeline
Trial start
The ClinicalTrials.gov record identifies March 2005 as the trial start.
Randomized treatment comparison
The trial is registered as phase 3, randomized, parallel, unmasked, with a primary purpose of treatment.
Primary completion
The ClinicalTrials.gov record identifies April 2008 as the primary completion date.
Outcome measures available
ClinicalTrials.gov is recorded as having posted results, with 13 outcome measures posted. The registry-reported statistical-analysis field remains empty.
16. Design Topics Not Reported in the ClinicalTrials.gov record
Several advanced trial-design issues are important in statistical analysis, but they should only be attributed to a specific trial when the trial record supports them. The registry-reported SYNTAX data does not provide details for the topics below.
| Design topic | Status in the ClinicalTrials.gov record | Why the distinction matters |
|---|---|---|
| Non-inferiority margin | Not reported | A non-inferiority margin is essential for interpreting a non-inferiority hypothesis and cannot be inferred from a randomized comparison alone. |
| Crossover | Not reported | Crossover can alter the relationship between randomized assignment and treatment actually received. |
| Factorial design | Not reported; design model is parallel | A factorial structure would create multiple intervention factors and interaction questions. |
| Interim analysis | Not reported | Repeated efficacy looks can require prespecified alpha-spending or group-sequential methods. |
| Missing-data / imputation strategy | Not reported | Binary endpoint results can depend on how participants with missing endpoint status are handled. |
| Stratification | Not reported | Stratified randomization and stratified analysis can affect the prespecified treatment comparison. |
| Bayesian methods | Not reported | A Bayesian analysis would require explicit prior and posterior modeling specifications. |
This distinction is not merely documentary. Each of these design choices can change how an effect estimate, uncertainty interval, or hypothesis test should be interpreted. The absence of a detail in the ClinicalTrials.gov recordset is therefore a reason not to reconstruct it from general knowledge.
17. Why This Trial Matters Statistically
SYNTAX is a useful statistical teaching case because its registry structure illustrates several recurring problems in clinical-trial analysis: randomized treatment comparison, binary composite endpoints, separate safety and clinical composites, multiple primary endpoints, and the importance of preserving denominators in safety reporting.
| Concept | How it appears in SYNTAX |
|---|---|
| Randomization | The trial is registered as randomized. |
| Parallel design | The registry specifies a parallel design with 2 arms. |
| Binary endpoint | All 3 registered primary endpoints are identified as binary. |
| Composite endpoint | MACCE combines death, stroke, documented myocardial infarction, and repeat revascularization. |
| Safety composite | The 12-month composite safety endpoint combines death, stroke, and documented myocardial infarction. |
| Separate revascularization endpoint | Repeat revascularization is itself a registered primary endpoint. |
| Multiplicity | Three primary endpoints create a multiple-hypothesis testing question. |
| Denominator awareness | Serious adverse events are reported as affected/at-risk counts by arm. |
| Statistical transparency | Results are posted, but no formal statistical-analysis entries are included in the ClinicalTrials.gov record. |
18. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The trial has a randomized parallel structure and three binary primary endpoints assessed at 12 months. Formal treatment-effect estimates and inferential statistics are not contained in the ClinicalTrials.gov record.
Clinical interpretation
The registered outcomes cover a broad clinical composite, a narrower safety composite, and repeat revascularization. These outcomes address different aspects of the treatment comparison and should not be collapsed into a single concept.
A statistically complete analysis would ultimately need to connect the endpoint definition to an effect measure, confidence interval, hypothesis test, analysis population, missing-data strategy, and multiplicity framework.
19. Important Limitations and Interpretation Issues
- No formal statistical-analysis entries: the ClinicalTrials.gov recordset contains no formal statistical analyses, so no primary-endpoint effect estimates, confidence intervals, or p-values can be reported from the ClinicalTrials.gov record.
- Posted outcomes without numerical values: ClinicalTrials.gov is identified as having results posted and 13 outcome measures posted, but the ClinicalTrials.gov record does not provide numerical values for the three primary outcomes.
- Composite interpretation: MACCE contains four different event categories. A composite result should not automatically be interpreted as showing an identical effect on every component.
- Multiple primary endpoints: three primary endpoints create an inferential multiplicity issue. The ClinicalTrials.gov record does not report the procedure used to address it.
- Missing-data uncertainty: the ClinicalTrials.gov record does not describe how missing 12-month endpoint information was handled.
- Analysis-population uncertainty: the ClinicalTrials.gov record does not define the formal efficacy or safety analysis populations beyond the provided serious-adverse-event denominators.
- Unmasked design: the registry records masking as none. Lack of masking can be relevant to outcome assessment and treatment processes, although the ClinicalTrials.gov record does not quantify any resulting bias.
- Unreported design details: non-inferiority margin, crossover, interim analysis, stratification, imputation, and Bayesian methods are not specified in the ClinicalTrials.gov record.
- Denominator interpretation: safety counts should be read together with their reported at-risk denominators rather than as standalone event counts.
20. What a Complete Binary-Endpoint Analysis Would Report
For an educational statistical workflow, a complete report of each SYNTAX primary endpoint would ideally include the event count and denominator in each randomized group, the corresponding event proportions, an absolute effect measure such as risk difference, a relative effect measure such as risk ratio or odds ratio, a 95% confidence interval, and the prespecified hypothesis-test result.
| Reporting element | Purpose |
|---|---|
| Events / denominator | Shows how many participants experienced the endpoint and the population from which the proportion is calculated. |
| Event proportion | Describes the absolute frequency of the binary outcome within each treatment group. |
| Risk difference | Shows the absolute separation between the two event proportions. |
| Risk ratio | Shows the event probability in one group relative to the other. |
| Odds ratio | Compares the odds of the binary endpoint between treatment groups. |
| 95% confidence interval | Describes uncertainty around the chosen effect estimate under the selected statistical framework. |
| P-value | Addresses the prespecified null hypothesis; it does not measure effect size. |
| Multiplicity adjustment | Places the result in the context of the three-primary-endpoint testing structure. |
This framework also illustrates why a p-value should never be presented alone. A statistically significant result without an absolute event rate or effect estimate gives little information about magnitude. Conversely, an effect estimate without its confidence interval does not adequately communicate statistical precision.
21. Confidence Intervals and Effect Size
A binary endpoint comparison should be described using an effect measure rather than only a significance test. A risk difference answers an absolute question: how much higher or lower is the event proportion in one group? A risk ratio answers a relative question: how many times as frequent is the event? These are complementary descriptions rather than interchangeable statistics.
A 95% confidence interval would quantify uncertainty around the selected effect estimate under the corresponding statistical framework. A narrow interval indicates greater statistical precision than a wide interval, but neither interval width nor the p-value alone determines clinical importance.
A p-value describes compatibility with a null hypothesis under the chosen test. It does not state how large the treatment difference is, how many patients benefit, or whether a result is clinically important. Those questions require the effect estimate and its absolute event rates.
22. Clinical Biostats Interpretation of the Evidence Structure
The most important statistical feature of the registry-reported SYNTAX record is the distinction between registered outcomes and available formal analyses. The registry identifies three primary binary endpoints and confirms that results are posted. However, the registry-reported statistical-analysis field contains no entries.
That distinction prevents a common reporting error: taking a trial with posted outcomes and assuming that every conventional statistical result is available for reconstruction. A rigorous analysis should report only estimates that are actually documented or can be derived directly from the ClinicalTrials.gov record under an explicitly stated calculation.
For SYNTAX, the appropriate educational focus is therefore the statistical architecture of the comparison: randomized allocation, two parallel treatment strategies, 12-month binary endpoint definitions, composite construction, multiple primary endpoints, and arm-specific serious-adverse-event denominators.
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Statistical Calculators
25. Sources
- ClinicalTrials.gov: NCT00114972.
- PubMed record: PMID 39447048.
- PubMed record: PMID 38465592.
- PubMed record: PMID 37993203.
- PubMed record: PMID 37265868.
- PubMed record: PMID 37256598.
Continue through the Clinical Biostats statistical pathway
Explore the broader methods behind randomized clinical-trial design, binary endpoints, treatment comparisons, and statistical interpretation.
26. Record Summary
SYNTAX is registered as a phase 3, randomized, parallel-group treatment trial with 1800.0 participants and 2 arms, comparing a polymer-based paclitaxel-eluting TAXUS Express2-SR stent with coronary artery bypass surgery in coronary artery disease. Its three registered primary endpoints are binary outcomes assessed at 12 months: 12-Month Binary MACCE, the 12-month composite safety endpoint, and repeat revascularization.
The statistical teaching value of the record lies in understanding how these binary and composite endpoints should be analysed and interpreted. MACCE combines death, stroke, documented myocardial infarction, and repeat revascularization, while the safety composite excludes repeat revascularization. The presence of three primary endpoints also creates a multiplicity question that requires a prespecified inferential framework.
The ClinicalTrials.gov record reports serious adverse events as 391/897 for CABG and 357/903 for PCI With DES. It also confirms that results are posted and that 13 outcome measures are posted.