This page separates registry-reported trial information from statistical interpretation. The ClinicalTrials.gov record identifies the study design, population, interventions, enrollment, sponsor, and status, but does not report registered primary endpoints or posted statistical analyses.
1. Trial at a Glance
ASTEC was a randomized phase 3 treatment trial evaluating surgery with or without lymphadenectomy and radiation therapy in patients with endometrial cancer. The registry records an enrollment of 2300 participants and identifies the study as completed.
| Feature | ASTEC |
|---|---|
| Study acronym | ASTEC |
| Phase | Phase 3 |
| Status | Completed |
| Condition | Endometrial Cancer |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 2300 |
| Start date | 1998-04 |
| Lead sponsor | Medical Research Council |
| Sponsor type | OTHER_GOV |
| ClinicalTrials.gov | NCT00003749 |
2. Clinical Question
The trial addresses a treatment question in endometrial cancer: how outcomes should be evaluated when surgical management includes conventional surgery, with the study examining surgery with or without lymphadenectomy and radiation therapy.
Population
Patients with endometrial cancer.
Intervention framework
The registry identifies adjuvant therapy, conventional surgery, brachytherapy, and radiation therapy among the trial interventions.
Comparator framework
The study title explicitly describes surgery with or without lymphadenectomy and radiation therapy, indicating that the randomized comparison concerns alternative treatment strategies involving these procedures and therapies.
Primary question
What is the effect of the randomized treatment strategies evaluated in ASTEC among patients with endometrial cancer?
3. Trial Design
Randomization is the central statistical feature of this design. When treatment is assigned randomly, the treatment groups can be compared as randomized rather than relying solely on adjustment for observed baseline characteristics. The resulting comparison is the basis for estimating the effect of the randomized treatment strategy.
The registry identifies ASTEC as a phase 3 study with treatment as its primary purpose. It began in 1998-04 and is recorded as completed.
4. Interventions
Conventional surgery
- Conventional surgery is listed as a trial intervention.
- The study title specifies surgery with or without lymphadenectomy.
Radiation therapy
- Radiation therapy is listed as a trial intervention.
- Brachytherapy is separately listed as a radiation intervention.
- Adjuvant therapy is also identified among the interventions.
The registry's intervention list therefore describes a treatment program involving surgery and radiation-related strategies. The ClinicalTrials.gov record does not provide treatment schedules, doses, procedural timing, or arm-specific sample sizes.
5. Endpoints
The ClinicalTrials.gov record does not report any registered primary endpoints for ASTEC. Consequently, endpoint names, operational definitions, assessment schedules, and prespecified analysis time frames are not available from the registry record.
| Endpoint information | Registry record |
|---|---|
| Registered primary endpoints | None reported |
| Endpoint definitions | Not reported in the registry record |
| Endpoint time frames | Not reported in the registry record |
| Posted statistical analyses | None reported |
This distinction is statistically important. A clinical trial can have a detailed protocol and statistical analysis plan even when the registry record available for review does not contain the corresponding endpoint definitions. Without those definitions, an endpoint-specific interpretation cannot be attributed to the registry.
6. Statistical Methodology
Because the registry does not post statistical analyses or registered primary endpoint definitions, the statistical methodology that would ordinarily be appropriate depends on the actual endpoint being evaluated. The appropriate analysis is determined by the endpoint's measurement scale, follow-up structure, censoring pattern, and prespecified estimand.
Binary endpoints
If an outcome is defined as the occurrence or nonoccurrence of a clinical event over a specified assessment period, a randomized comparison could use an appropriate risk measure such as a risk difference, risk ratio, or odds ratio. Confidence intervals should accompany the point estimate so that the uncertainty of the treatment comparison is visible.
Continuous endpoints
If an endpoint is measured on a continuous scale, a treatment-group difference can be estimated using a suitable comparison of means or a regression model. If baseline measurements are available, analysis of covariance can improve precision by accounting for the baseline value when that approach is prespecified and appropriate.
Time-to-event endpoints
If an outcome measures time from randomization to an event, survival-analysis methods are generally appropriate. Kaplan-Meier estimation can describe the event-time distribution, while a log-rank test can compare randomized groups. A Cox proportional-hazards model can estimate a hazard ratio when its modeling assumptions are appropriate.
Time-to-event analysis accommodates right-censored observations, allowing participants who have not experienced the event by their last known follow-up to contribute information up to the censoring time.
Count or rate endpoints
If the outcome is a count of events or a rate accumulated over follow-up, a Poisson or negative-binomial model may be appropriate depending on the event distribution and degree of overdispersion. The model should reflect the actual estimand and observation time rather than treating all participants as if they had identical follow-up.
Adjusted analyses
Regression adjustment can account for prespecified prognostic variables and potentially improve precision. In a randomized trial, however, adjustment should not replace the primary randomized comparison. Covariates should be selected according to the protocol or statistical analysis plan rather than chosen after looking at treatment-group differences.
7. Statistical Methods Explained
Why does randomization matter?
Randomization creates the foundation for a causal comparison between treatment strategies. Under proper implementation, treatment assignment is independent of baseline prognostic factors in expectation. This allows differences in outcomes between randomized groups to be attributed to the assigned treatment strategy under the trial's assumptions.
Why must the endpoint be defined before choosing the statistical test?
Statistical methods are determined by the structure of the outcome. A binary endpoint, continuous measurement, count, and time-to-event outcome contain different information and require different probability models. Selecting the analysis after observing the results can change the interpretation of statistical evidence.
When would a Kaplan-Meier analysis be appropriate?
Kaplan-Meier estimation is useful when the endpoint is time to an event and some participants have incomplete follow-up. It estimates the probability of remaining event-free over time while accounting for right censoring.
What does a hazard ratio measure?
A hazard ratio is a relative measure comparing the instantaneous event rates between treatment groups within a survival model. A value below one indicates a lower modeled event hazard for the numerator group, whereas a value above one indicates a higher modeled event hazard. It is not the same as a risk ratio and does not directly represent an absolute difference in event probability.
Why are confidence intervals important?
A point estimate alone does not show how precisely the treatment effect has been estimated. A confidence interval describes uncertainty under the specified statistical model and sampling framework. Wider intervals generally indicate less statistical precision than narrower intervals.
Why should analysis populations be specified?
Randomized efficacy analyses generally preserve treatment assignment through an intention-to-treat framework, while safety analyses can be based on participants who actually received treatment. Different analysis populations answer different questions, so the population used for an estimate should always be stated.
8. Planned Analysis
The ASTEC registry record identifies a randomized phase 3 treatment trial involving patients with endometrial cancer and interventions including conventional surgery, adjuvant therapy, brachytherapy, and radiation therapy. It does not report registered primary endpoint definitions or posted statistical analyses.
For a randomized treatment comparison, the primary analysis would ordinarily compare outcomes according to randomized assignment using the statistical method appropriate to the prespecified endpoint. If the endpoint were a time-to-event outcome, a Kaplan-Meier analysis with a treatment-group comparison such as a log-rank test and an effect estimate from a survival model would ordinarily be considered. If the endpoint were binary, an appropriate comparison of event proportions with a confidence interval would ordinarily be used. Continuous outcomes would generally require a difference or regression-based estimate appropriate to the measurement scale.
The distinction between a planned statistical method and an observed result is fundamental. A method describes how evidence would be summarized and tested; it does not establish the direction or magnitude of an outcome.
9. Randomization and Causal Interpretation
The registry identifies ASTEC as randomized. From a statistical perspective, this is more consequential than simply labeling the study as comparative. Randomization determines the treatment assignment mechanism and provides the basis for comparing potential outcomes across treatment strategies.
What randomization provides
It provides a principled mechanism for balancing both measured and unmeasured baseline factors in expectation across randomized groups.
What randomization does not guarantee
It does not guarantee identical baseline characteristics in a finite sample, complete follow-up, perfect adherence, or absence of missing observations.
What the analysis must preserve
The primary efficacy analysis should respect the randomized assignment and the estimand specified for the trial.
Why estimands matter
The treatment effect depends on the precise clinical question, including how treatment discontinuation, additional treatment, and other post-randomization events are handled.
10. Analysis of Treatment Strategies
The title describes surgery with or without lymphadenectomy and radiation therapy, while the intervention list includes conventional surgery, brachytherapy, radiation therapy, and adjuvant therapy. This makes treatment-strategy definition particularly important when interpreting any eventual efficacy analysis.
A treatment comparison should be tied to the actual randomized assignment rather than retrospectively dividing participants according to procedures they eventually received. Post-randomization treatment changes can be clinically meaningful, but analyzing patients according to treatment actually received can introduce selection bias because treatment received after randomization may depend on prognosis or other evolving factors.
| Statistical issue | Why it matters |
|---|---|
| Randomized assignment | Defines the primary causal comparison. |
| Procedure received | May differ from randomized assignment and can be influenced by post-randomization information. |
| Radiation exposure | Should be interpreted according to the prespecified treatment strategy and timing. |
| Lymphadenectomy | The study title specifically identifies its presence or absence as part of the treatment comparison. |
| Adjuvant therapy | The registry lists adjuvant therapy among the interventions and therefore treatment exposure should be defined precisely in any endpoint analysis. |
11. Statistical Interpretation of Treatment Effects
A treatment-effect estimate based on randomized assignment answers a different question from an association based on treatments actually received. The randomized comparison is protected by the treatment assignment mechanism; a post-randomization comparison can be affected by factors that influence which treatment a participant receives.
Relative measures can describe the proportional difference between groups, while absolute measures describe the difference in outcome probability or outcome level. Both can be useful, but neither should be substituted for the other without considering the clinical endpoint and time frame.
Every treatment estimate should be interpreted together with its uncertainty interval. The width of that interval reflects statistical precision and is influenced by factors such as sample size, event frequency, variability, and study design.
12. Limitations
- Endpoint information: The ClinicalTrials.gov record does not report registered primary endpoints or their operational definitions.
- Statistical analyses: No posted statistical analyses are reported in the registry record.
- Outcome interpretation: Without endpoint-specific definitions and estimates, a treatment effect cannot be characterized from the registry record alone.
- Analysis population: The registry information available here does not provide arm-specific analysis populations or definitions of efficacy and safety populations.
- Treatment detail: The intervention list does not provide the complete treatment schedules, procedural timing, or exposure details required for a detailed per-protocol treatment analysis.
- Missing-data methods: The registry record does not specify missing-data or imputation procedures.
- Multiplicity: The registry record does not describe multiplicity adjustments, endpoint hierarchy, or allocation of type I error.
- Interim monitoring: The registry record does not describe interim analyses or stopping boundaries.
- Stratification: The registry record does not identify randomization stratification factors.
- Subgroups: The registry record does not provide prespecified subgroup analyses or interaction tests.
13. Why This Trial Matters Statistically
ASTEC is a useful statistical teaching case because it illustrates an important distinction between trial design information and reported statistical evidence. A registry can establish that a study was randomized, phase 3, completed, and designed for treatment evaluation while providing little or no endpoint-level statistical information.
| Concept | How it appears in ASTEC |
|---|---|
| Randomization | The registry identifies the allocation as randomized. |
| Phase 3 design | The study is registered as phase 3. |
| Treatment purpose | The primary purpose is recorded as treatment. |
| Large enrollment | The registry records enrollment of 2300 participants. |
| Intervention complexity | The intervention list includes conventional surgery, adjuvant therapy, brachytherapy, and radiation therapy. |
| Endpoint specification | Registered primary endpoints are not reported in the registry record. |
| Statistical transparency | No posted statistical analyses are reported in the registry record. |
| Estimand definition | The available registry information does not specify the endpoint-specific estimand. |
| Analysis population | The registry information does not identify endpoint-specific analysis populations. |
For statisticians, this is a useful reminder that the words randomized and phase 3 describe important design properties but do not, by themselves, specify the correct statistical analysis. The endpoint, estimand, analysis population, censoring rules, treatment deviations, and multiplicity framework must also be established.
14. Statistical Reasoning for a Randomized Surgical Trial
Surgical trials can present statistical challenges that differ from trials involving a single continuously administered drug. Procedures can be modified, omitted, or combined with other treatment components, and the clinical pathway can involve several interventions.
Assignment is not exposure
The randomized treatment assignment is determined at randomization, whereas actual treatment exposure can be affected by clinical circumstances after assignment.
Composite strategies need definition
When several interventions are included in a treatment strategy, the estimand should make clear whether the effect concerns the complete strategy or an individual component.
Time matters
If treatment components occur at different points during follow-up, the timing of treatment and outcome assessment can affect the statistical estimand.
Missing follow-up matters
Incomplete follow-up can affect both precision and validity, particularly for time-to-event outcomes when censoring is related to prognosis.
15. What a Complete Analysis Would Need to Establish
A full statistical analysis of ASTEC would require more information than is contained in the registry record. In particular, endpoint definitions would determine the analysis model, while treatment assignment and follow-up rules would determine the appropriate analysis population and handling of censoring.
| Required statistical element | Purpose |
|---|---|
| Primary endpoint definition | Defines the outcome being tested. |
| Time frame | Defines when the outcome is assessed or how follow-up is measured. |
| Estimand | Defines the treatment effect being targeted. |
| Analysis population | Defines which randomized participants contribute to the primary estimate. |
| Missing-data strategy | Defines how incomplete observations are handled. |
| Statistical model | Defines how the treatment effect is estimated. |
| Multiplicity framework | Defines how multiple hypotheses or endpoints affect type I error. |
| Interim-analysis plan | Defines how repeated examination of accumulating data is incorporated into inference. |
These elements are not interchangeable. For example, a confidence interval can quantify uncertainty around an estimate, but it cannot define the clinical question being estimated. Likewise, randomization establishes the treatment assignment mechanism, but it does not determine whether a binary, continuous, or time-to-event model is appropriate.
16. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The available registry information establishes a randomized phase 3 treatment design with 2300 participants but does not provide endpoint-specific estimates or posted statistical analyses.
Clinical interpretation
The study concerns treatment strategies for endometrial cancer involving surgery, lymphadenectomy, and radiation-related interventions. The clinical meaning of those strategies requires endpoint-specific outcome evidence.
Keeping these two levels separate prevents a common statistical error: treating a study design characteristic as if it were an outcome. Randomization describes how treatment was assigned; it does not establish whether one strategy produced better or worse outcomes.
17. Planned Analysis by Endpoint Type
Because the registry does not identify its registered primary endpoints, the following framework describes how common endpoint structures would ordinarily be analyzed in a randomized phase 3 treatment trial.
| Endpoint structure | Typical primary analysis | Common effect measure |
|---|---|---|
| Binary outcome | Comparison of event proportions or regression model | Risk difference, risk ratio, or odds ratio |
| Continuous outcome | Comparison of means or covariance-adjusted regression | Mean difference |
| Time-to-event outcome | Kaplan-Meier and appropriate survival-model comparison | Hazard ratio or time-specific survival difference |
| Count outcome | Poisson or negative-binomial modeling when appropriate | Rate ratio |
The correct choice among these methods should be determined before examining treatment-group outcome differences. The statistical method is part of the definition of the evidence, not merely a mathematical presentation added after the result is known.
18. Missing Data and Censoring
The registry record does not specify a missing-data or imputation strategy. That absence is particularly relevant if any primary endpoint requires longitudinal follow-up.
For a continuous endpoint, missing observations can change the effective sample size and potentially introduce bias if missingness is related to treatment or prognosis. Multiple imputation, mixed-effects models, or other approaches may be appropriate in particular settings, but their validity depends on assumptions about the missing-data mechanism and the estimand.
For a time-to-event endpoint, incomplete follow-up is generally represented through censoring rather than simple deletion. The validity of a survival analysis depends in part on whether the censoring mechanism is appropriately handled and whether the assumptions underlying the analysis are reasonable.
Censoring is not the same as an event. A participant who is censored contributes information through the time at which follow-up ends, but is not treated as having experienced the event merely because observation stopped.
19. Multiplicity and Interim Analysis
The registry does not describe a multiplicity strategy or interim-analysis plan for ASTEC. These design features are important whenever a trial has multiple confirmatory endpoints, repeated opportunities to evaluate efficacy, or several formally tested hypotheses.
If multiple hypotheses are tested at the same significance level without adjustment, the probability of at least one false-positive finding can exceed the nominal level for an individual test. Prespecified multiplicity procedures can control the relevant familywise error rate while allowing several clinically meaningful questions to be addressed.
Similarly, repeated unplanned inspection of accumulating efficacy data can affect statistical inference. Group-sequential methods and alpha-spending approaches can incorporate planned interim looks while maintaining the desired type I error properties.
The number and relationship of confirmatory hypotheses, together with any interim analyses, determine how statistical evidence should be interpreted.
20. Subgroups and Effect Modification
The registry information available for ASTEC does not report prespecified subgroup analyses or interaction tests. Subgroup analysis is nevertheless an important statistical concept for randomized trials because an overall treatment effect does not automatically imply that the effect is identical across every patient characteristic.
A subgroup-specific estimate can differ from the overall estimate because of sampling variability alone. To evaluate whether treatment effect truly differs between subgroups, the appropriate statistical question is an interaction or treatment-by-subgroup effect, rather than whether one subgroup has a statistically significant result while another does not.
21. Bayesian Methods
The registry record does not describe Bayesian statistical methods for ASTEC. A Bayesian analysis would require explicit specification of prior distributions, a likelihood, the estimand, and the posterior quantity used for inference.
Bayesian and frequentist analyses can address similar clinical questions, but they express uncertainty differently. A posterior probability is conditional on the specified prior and observed data, whereas a frequentist confidence interval has a different repeated-sampling interpretation.
22. Longitudinal Trial History
Trial begins
The ClinicalTrials.gov record lists 1998-04 as the study start date.
Randomized treatment study
The registry identifies ASTEC as a phase 3 randomized treatment trial in endometrial cancer.
Study status
The registry records the study as completed.
The available registry information establishes the broad design history but does not provide a sequence of interim efficacy analyses, final statistical analyses, or endpoint-specific result updates.
23. Statistical Concepts in This Trial
Learn more about the statistical ideas that are useful when interpreting randomized clinical trials:
24. Statistical Methods and Clinical Trial Design
ASTEC illustrates why a rigorous clinical-trial analysis begins with the estimand and endpoint rather than with a preferred statistical test. The randomized design supplies the comparison framework, but the endpoint determines the appropriate statistical model.
| Design element | Statistical question |
|---|---|
| Randomized allocation | What treatment effect is estimated by preserving randomized assignment? |
| Phase 3 | What confirmatory evidence was prespecified for the primary clinical question? |
| Multiple interventions | Is the estimand the effect of a complete treatment strategy or an individual component? |
| Lymphadenectomy | How should the randomized strategy involving its presence or absence be defined? |
| Radiation therapy | How should the radiation component be incorporated into the treatment estimand? |
| Follow-up | How should incomplete observation and timing of events be handled? |
25. Sources
- ClinicalTrials.gov: ASTEC, NCT00003749.
Continue through the Clinical Biostats knowledge graph
Explore broader statistical tutorials, calculators, and clinical-trial analyses to deepen understanding of randomized study design and analysis.
26. Record Summary
ASTEC is registered as a completed, randomized phase 3 treatment trial involving 2300 participants with endometrial cancer. The registry identifies a treatment strategy involving surgery with or without lymphadenectomy and radiation therapy and lists adjuvant therapy, conventional surgery, brachytherapy, and radiation therapy among the interventions.
From a statistical perspective, the most important feature of the available record is the distinction between a clearly identified randomized trial design and the absence of posted endpoint-level statistical analyses. A rigorous interpretation requires the primary endpoint definition, time frame, estimand, analysis population, treatment comparison, missing-data strategy, and prespecified inferential framework before numerical treatment effects can be interpreted.