This page separates reported trial results from statistical interpretation. Numerical results are taken from the ClinicalTrials.gov record. The analysis focuses on the registered endpoints and statistical analyses posted for TEXT.
1. Trial at a Glance
TEXT was a randomized, parallel-group phase 3 oncology trial comparing triptorelin with tamoxifen against triptorelin with exemestane in women with breast cancer. The registry identifies disease-free survival as the single primary endpoint and reports four statistical analyses using log-rank testing and hazard ratios.
| Feature | TEXT |
|---|---|
| Trial name | TEXT |
| NCT identifier | NCT00066703 |
| Phase | Phase 3 |
| Status | Completed |
| Condition | Breast Cancer |
| Population | Premenopausal women with hormone-responsive breast cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 2672 |
| Lead sponsor | ETOP IBCSG Partners Foundation |
2. Clinical Question
The central statistical question was whether the randomized treatment comparison between triptorelin plus tamoxifen (T+OFS) and triptorelin plus exemestane (E+OFS) was associated with a difference in disease-free survival in premenopausal women with hormone-responsive breast cancer.
Population
Premenopausal women with hormone-responsive breast cancer, as described by the trial's brief title.
Intervention
Triptorelin with exemestane, identified in the posted analyses as the E+OFS group.
Comparator
Triptorelin with tamoxifen, identified in the posted analyses as the T+OFS reference group.
Primary question
Does the E+OFS versus T+OFS randomized comparison differ in disease-free survival?
The registry identifies the hypothesis type as superiority. This means the statistical framework was designed to test whether the randomized groups differed in the direction represented by the prespecified treatment comparison, rather than to establish that the two strategies were sufficiently similar.
3. Trial Design
Triptorelin + tamoxifen
- Triptorelin
- Tamoxifen
- Reference group for the posted hazard-ratio estimates
Triptorelin + exemestane
- Triptorelin
- Exemestane
- Compared with the T+OFS reference group
The registry does not provide, in the ClinicalTrials.gov record, details on stratification factors, crossover, interim-analysis boundaries, multiplicity procedures, missing-data imputation, or Bayesian methods. Those design features therefore are not assumed here.
4. Trial Timeline
Trial start
The registry records a study start date of 2003-11-03.
Primary completion
The registry records primary completion on 2011-03-11.
Disease-free survival analysis
The primary disease-free survival result is a 5-year estimate reported at a median follow-up of 72 months.
Overall survival analysis
The posted overall-survival analysis reports 8-year estimates at a median follow-up of 9 years.
5. Endpoints
| Endpoint | Role | Time frame | Registry definition / description |
|---|---|---|---|
| Disease-free Survival | Primary | 5-year estimate reported at a median follow-up of 72 months | Estimated percentage of patients alive and disease-free at 5 years from randomization, where disease-free survival is defined as the time from randomization to the first appearance of one of the following: invasive breast cancer recurrence at local, regional, or distant site, invasive contralateral breast cancer, second (non-breast) invasive cancer, or death without cancer event; or censored at da |
| Breast Cancer-free Interval | Secondary | 5-year estimate reported at a median follow-up of 72 months | Time-to-event endpoint analyzed in the intention-to-treat population. |
| Distant Recurrence-free Interval | Secondary | 5-year estimates reported at a median follow-up of 72 months | Time-to-event endpoint analyzed in the intention-to-treat population. |
| Overall Survival | Secondary | 8-year estimates, reported at a median follow-up of 9 years | Time-to-event endpoint analyzed in the intention-to-treat population. |
6. Statistical Methodology
Intention-to-treat analysis
The posted analyses use the intention-to-treat (ITT) population. In an ITT analysis, participants are analyzed according to the treatment group to which they were randomized. The important statistical principle is that randomization defines the comparison being estimated.
For a randomized trial, this approach protects the treatment comparison from being redefined after randomization based on subsequent treatment exposure, adherence, or other post-randomization events. The ITT framework therefore keeps the primary comparison anchored to the original randomized groups.
Log-rank test
The registry reports Log Rank as the method for all four statistical analyses. The log-rank test is designed for comparing survival or other time-to-event distributions between groups while accounting for the timing of events and right censoring.
The log-rank test evaluates evidence of a difference in the time-to-event experience between groups. It is not itself an estimate of how large that difference is.
Hazard ratio
The reported effect measure for every posted analysis is a hazard ratio. The registry explicitly identifies T+OFS as the reference group in the estimation of the hazard ratio.
Under this reference-group convention, an HR below 1 indicates a lower estimated instantaneous event rate in E+OFS relative to T+OFS within the fitted time-to-event comparison.
The hazard ratio is a relative time-to-event measure. It does not directly tell us the absolute percentage of participants who experienced an event, the absolute difference in event-free survival at a particular time point, or the probability that an individual participant benefits.
Confidence intervals
The primary disease-free survival analysis reports a two-sided 95% confidence interval of 0.602 to 0.855 around the hazard-ratio estimate of 0.717. A confidence interval provides information about the precision of the estimated treatment effect under the statistical model and sampling framework.
The interval is also useful because it displays more information than a single point estimate. The estimate of 0.717 is the center of the reported treatment-effect summary, while the interval communicates the statistical uncertainty surrounding that estimate.
P-values
The primary disease-free survival analysis reports P = .0002. The p-value quantifies the compatibility of the observed data with the null hypothesis used by the statistical test. It is not a measure of effect size and should not be read as the probability that the null hypothesis is true.
Right censoring and time-to-event analysis
Time-to-event methods are needed because participants may have different amounts of observed follow-up and may not all experience the endpoint during the observation period. A participant whose event has not occurred by the relevant observation time can contribute information without being treated as though an event occurred.
7. Primary Result: Disease-free Survival
The primary endpoint is disease-free survival, with a 5-year estimate reported at a median follow-up of 72 months. The analysis was conducted in the intention-to-treat population using a log-rank test, with the hazard ratio estimated using T+OFS as the reference group.
Disease-free survival hazard ratio
95% CI: 0.602–0.855 · P = .0002
5-year estimate reported at a median follow-up of 72 months
| Primary endpoint | Analysis population | Method | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|---|
| Disease-free Survival | Intention-to-treat | Log Rank | Hazard Ratio | 0.717 | 0.602–0.855 | .0002 |
The reported HR of 0.717 means that, under the hazard-ratio framework and with T+OFS as the reference group, the estimated instantaneous event rate for the E+OFS group was approximately 71.7% of the corresponding rate for T+OFS. Expressed as a simple relative-hazard interpretation, this corresponds to an estimated 28.3% lower hazard because 1 − 0.717 = 0.283.
The HR does not mean that 28.3% of participants avoided an event, that an individual participant had exactly a 28.3% lower probability of recurrence or death, or that the absolute 5-year difference between the groups was 28.3 percentage points. Those are different quantities.
The 95% CI of 0.602–0.855 communicates the precision of the hazard-ratio estimate. It describes uncertainty around the estimated relative hazard under the statistical framework; it does not describe the range of individual treatment effects among patients.
The P = .0002 result is evidence against the null hypothesis evaluated by the log-rank analysis. It should not be interpreted as a 0.02% probability that the treatment groups are identical, nor does it quantify the magnitude or clinical importance of the hazard-ratio estimate.
Because this is a time-to-event analysis, interpretation also depends on the handling of censoring and on the assumptions underlying the hazard-ratio representation. The ClinicalTrials.gov record does not provide enough information to independently assess the proportional-hazards assumption.
What the primary result establishes statistically
The registry reports a formal superiority analysis of disease-free survival with a hazard ratio below 1, a two-sided 95% confidence interval, and a p-value of .0002. Thus, the posted analysis provides both an estimated relative treatment effect and an inferential test of the randomized comparison.
The result should nevertheless be kept conceptually separate from an absolute-risk statement. A hazard ratio describes relative event rates over time, while a 5-year survival percentage would describe the estimated proportion remaining event-free at a specific time point. The registry-reported statistical analysis reports the 5-year time frame but does not provide the corresponding arm-specific percentages in the posted analysis data.
8. Secondary Result: Breast Cancer-free Interval
The breast cancer-free interval was analyzed as a secondary time-to-event endpoint using the same general statistical framework: intention-to-treat analysis, log-rank testing, and a hazard ratio with T+OFS as the reference group.
Breast cancer-free interval hazard ratio
95% CI: .548–.804 · P < .0001
5-year estimate reported at a median follow-up of 72 months
| Endpoint | Method | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Breast Cancer-free Interval | Log Rank; ITT | HR 0.664 | .548–.804 | <.0001 |
The HR of 0.664 indicates an estimated instantaneous event rate approximately 66.4% as large in E+OFS as in T+OFS under the reported hazard-ratio framework. The corresponding simple relative-hazard interpretation is an estimated 33.6% lower hazard because 1 − 0.664 = 0.336.
The 95% CI of .548–.804 indicates uncertainty around that estimate. It does not mean that individual participants' treatment effects fall inside that interval.
The P < .0001 result provides evidence against the null hypothesis evaluated by the log-rank test. It does not indicate that the treatment effect is "more than" some particular clinical threshold, and it should not be used as a substitute for the effect estimate and confidence interval.
Because this is a secondary endpoint, its inferential interpretation should remain distinct from the primary disease-free survival analysis. The ClinicalTrials.gov record identifies the hypothesis type as superiority but does not provide additional multiplicity-adjustment details for this secondary endpoint.
9. Secondary Result: Distant Recurrence-free Interval
Distant recurrence-free interval was also analyzed as a secondary time-to-event endpoint. The registry reports a 5-year estimate at a median follow-up of 72 months, with intention-to-treat analysis and a log-rank test.
Distant recurrence-free interval hazard ratio
95% CI: 0.624–0.967 · P = 0.02
5-year estimates reported at a median follow-up of 72 months
| Endpoint | Method | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Distant Recurrence-free Interval | Log Rank; ITT | HR 0.777 | 0.624–0.967 | 0.02 |
The HR of 0.777 corresponds to an estimated instantaneous event rate about 77.7% of that in the T+OFS reference group. As a simple relative-hazard transformation, 1 − 0.777 gives an estimated 22.3% lower hazard.
The 95% CI of 0.624–0.967 shows the uncertainty around the estimated hazard ratio. Because the interval remains below 1, the estimate is consistent with a lower estimated hazard in E+OFS under this model.
The P = 0.02 value is evidence against the null hypothesis specified for the log-rank comparison. It does not describe the size of the treatment effect. The hazard ratio and its confidence interval are the appropriate quantities for communicating magnitude and precision.
This endpoint is secondary, so its result should not be mentally substituted for the primary disease-free survival analysis. The statistical hierarchy matters when interpreting a collection of endpoint-specific p-values.
10. Secondary Result: Overall Survival
Overall survival was posted as a secondary endpoint. The registry reports an 8-year estimate at a median follow-up of 9 years, analyzed in the intention-to-treat population with a log-rank test and hazard ratio.
Overall survival hazard ratio
95% CI: 0.79–1.22 · P = 0.84
8-year estimates, reported at a median follow-up of 9 years
| Endpoint | Method | Estimate | 95% CI | P-value |
|---|---|---|---|---|
| Overall Survival | Log Rank; ITT | HR 0.98 | 0.79–1.22 | 0.84 |
An HR of 0.98 is close to 1. Under the reported hazard-ratio framework, the estimated instantaneous rate of death in E+OFS was approximately 98% of that in the T+OFS reference group. This point estimate alone does not establish equivalence or prove that the two strategies have identical survival.
The 95% CI of 0.79–1.22 is especially important here because it spans 1. The interval represents uncertainty around the estimated hazard ratio and includes values corresponding to lower and higher estimated hazards for E+OFS relative to T+OFS.
The P = 0.84 result does not mean that there is an 84% probability that the treatment effects are equal. Rather, it indicates that the observed data are not unusual under the null hypothesis evaluated by the statistical test. A non-significant p-value is not itself evidence of equivalence or non-inferiority.
The distinction matters because the trial's registered hypothesis type is superiority. A superiority analysis that does not reject its null hypothesis does not automatically establish that the two treatment strategies are statistically equivalent.
11. Putting the Four Analyses Together
The four posted analyses form a coherent time-to-event statistical set: one primary endpoint and three secondary endpoints. The hazard-ratio estimates are all below 1 except for overall survival, which is close to 1.
| Endpoint | Role | HR | 95% CI | P-value | Time frame |
|---|---|---|---|---|---|
| Disease-free Survival | Primary | 0.717 | 0.602–0.855 | .0002 | 5-year estimate; median follow-up 72 months |
| Breast Cancer-free Interval | Secondary | 0.664 | .548–.804 | <.0001 | 5-year estimate; median follow-up 72 months |
| Distant Recurrence-free Interval | Secondary | 0.777 | 0.624–0.967 | 0.02 | 5-year estimates; median follow-up 72 months |
| Overall Survival | Secondary | 0.98 | 0.79–1.22 | 0.84 | 8-year estimates; median follow-up 9 years |
These estimates should not be collapsed into a single "trial effect." Disease-free survival, breast cancer-free interval, distant recurrence-free interval, and overall survival represent different clinical event definitions. A treatment effect can differ across endpoints because each endpoint counts different events and therefore measures a different aspect of the patient's time-to-event experience.
The primary result and secondary results also illustrate why the effect estimate, confidence interval, and p-value should be read together. The HR communicates relative magnitude, the confidence interval communicates precision, and the p-value addresses compatibility with the statistical null hypothesis.
Relative effect
The hazard ratio summarizes the relative event rate between E+OFS and the T+OFS reference group.
Precision
The confidence interval shows the uncertainty around each reported hazard-ratio estimate.
Inference
The p-value summarizes evidence against the null hypothesis evaluated by the log-rank analysis.
Endpoint definition
The meaning of each hazard ratio depends on which event constitutes the endpoint.
12. Safety
The ClinicalTrials.gov record reports serious adverse events by randomized arm. The figures are presented as affected participants divided by participants at risk.
| Safety measure | T+OFS | E+OFS |
|---|---|---|
| Serious adverse events, affected / at risk | 484 / 1321 | 496 / 1317 |
The registry data does not provide a formal statistical comparison for serious adverse events in the ClinicalTrials.gov record. The affected/at-risk figures therefore should be treated as descriptive safety information rather than as evidence from a separate hypothesis test.
It is also important not to substitute these safety counts for efficacy results. Safety and efficacy answer different questions: the former concerns adverse outcomes associated with treatment exposure, while the latter concerns the prespecified disease-related time-to-event endpoints.
13. Statistical Methods Explained
Why was an intention-to-treat analysis used?
ITT preserves the randomized comparison. Once participants have been randomized, analyzing them according to their assigned group maintains the treatment contrast created by randomization. The TEXT registry explicitly identifies ITT as the analysis population for all four posted endpoint analyses.
Why use a log-rank test for disease-free survival?
Disease-free survival is a time-to-event endpoint rather than a simple binary outcome observed at one fixed time. The log-rank test uses the timing of events and accommodates censored observations, making it appropriate for comparing survival distributions between randomized groups.
What does a hazard ratio of 0.717 mean?
With T+OFS as the reference group, an HR of 0.717 means the estimated instantaneous event rate for E+OFS was approximately 71.7% of that in T+OFS under the hazard-ratio framework. It does not mean that exactly 28.3% fewer participants experienced an event.
Why is the confidence interval important?
A point estimate alone can give a false impression of precision. The 95% CI of 0.602–0.855 around the primary HR of 0.717 communicates uncertainty around the estimate. It is therefore more informative to report the HR and CI together than to focus on the HR alone.
Why does the p-value not measure effect size?
The p-value addresses evidence against a statistical null hypothesis. It depends on the observed data and the statistical test, whereas effect size is communicated by a measure such as the hazard ratio. A very small p-value can occur with a modest effect in a sufficiently informative dataset, while a clinically important estimate can have substantial uncertainty in a smaller or less informative dataset.
What does an overall-survival HR of 0.98 with P = 0.84 mean?
The point estimate is close to 1, and the confidence interval of 0.79–1.22 includes 1. The p-value does not provide evidence against the superiority null hypothesis in this analysis. It would be incorrect, however, to convert this into a claim of equivalence because the registry identifies the hypothesis type as superiority rather than non-inferiority or equivalence.
Why can different endpoints have different hazard ratios?
Each endpoint defines an event differently. Disease-free survival, breast cancer-free interval, distant recurrence-free interval, and overall survival therefore summarize different time-to-event processes. Their hazard ratios are not interchangeable, even though the same log-rank framework was used for the posted analyses.
14. Interpreting Hazard Ratios Carefully
The disease-free survival HR of 0.717 is a relative measure comparing the event rate in E+OFS with the T+OFS reference group. A convenient descriptive transformation is 1 − 0.717 = 0.283, or an estimated 28.3% lower hazard.
The HR does not directly provide the absolute probability of being disease-free at 5 years, the number needed to treat, the difference in median time to event, or an individual patient's probability of benefit. None of those quantities is reported in the ClinicalTrials.gov record.
A hazard ratio is a time-to-event quantity. It incorporates information from events occurring over follow-up rather than simply comparing two proportions at a single time point. The registry therefore reports the time frame alongside each analysis.
Hazard-ratio interpretation depends on the statistical model underlying the estimate. The ClinicalTrials.gov record identifies log-rank testing and hazard-ratio estimation but does not provide sufficient model details to independently assess whether proportional hazards held throughout follow-up.
15. Primary vs Secondary Endpoint Interpretation
The registry identifies disease-free survival as the single primary endpoint. Breast cancer-free interval, distant recurrence-free interval, and overall survival are secondary endpoints. This distinction is statistically important because an endpoint's role in the protocol determines how its result should be interpreted within the overall trial evidence.
| Endpoint role | Endpoint | Interpretive function |
|---|---|---|
| Primary | Disease-free Survival | Main prespecified efficacy comparison reported in the registry analysis. |
| Secondary | Breast Cancer-free Interval | Additional disease-related time-to-event comparison. |
| Secondary | Distant Recurrence-free Interval | Additional time-to-event comparison focused on distant recurrence. |
| Secondary | Overall Survival | Longer-term survival endpoint reported at 8 years. |
The ClinicalTrials.gov record does not specify a multiplicity-adjustment strategy for the secondary endpoints. Consequently, their p-values should be reported exactly as registered but should not be treated as though the ClinicalTrials.gov record established a particular familywise error-control procedure.
16. Randomization and Causal Interpretation
Randomization is one of the central statistical features of TEXT. By assigning participants randomly to the two treatment strategies, the trial creates a comparison in which treatment assignment is determined independently of participants' baseline characteristics in expectation.
The value of randomization is reflected in the ITT analysis. Rather than selecting patients after randomization based on treatment received or subsequent events, the analysis preserves the randomized groups. This allows the reported hazard-ratio comparisons to be interpreted as comparisons associated with randomized treatment assignment, subject to the usual assumptions of randomized-trial inference.
Randomization
Defines the treatment groups before post-randomization outcomes occur.
ITT
Maintains participants in their randomized groups for efficacy analysis.
Time-to-event analysis
Uses the timing of events and censoring rather than only whether an event occurred.
Hazard ratio
Summarizes the relative event rate between the two randomized groups.
17. What the Confidence Intervals Tell Us
The four posted analyses provide a useful illustration of how confidence intervals can change the interpretation of a point estimate.
| Endpoint | HR | 95% CI | Interpretive feature |
|---|---|---|---|
| Disease-free Survival | 0.717 | 0.602–0.855 | Interval lies below 1. |
| Breast Cancer-free Interval | 0.664 | .548–.804 | Interval lies below 1. |
| Distant Recurrence-free Interval | 0.777 | 0.624–0.967 | Interval lies below 1 but is closer to 1 at its upper bound. |
| Overall Survival | 0.98 | 0.79–1.22 | Interval includes 1. |
The confidence intervals should be read as measures of uncertainty rather than as ranges of possible outcomes for individual patients. They also should not be interpreted as a probability distribution for the true hazard ratio.
The contrast between distant recurrence-free interval and overall survival is particularly useful educationally. The distant recurrence-free interval estimate is below 1 and its reported interval is below 1, whereas the overall-survival interval crosses 1. The endpoints measure different events, so their statistical results need not move in parallel.
18. Overall Survival and the Meaning of a Non-Significant Result
The overall-survival analysis provides an important lesson in statistical interpretation. Its HR is 0.98, with a 95% CI of 0.79–1.22 and P = 0.84.
A common error is to describe a non-significant superiority result as proving that the treatments are the same. That is not what the analysis establishes. Failure to reject a superiority null hypothesis is compatible with several possibilities, including little difference and a difference that the available information did not estimate precisely enough to distinguish from the null.
The confidence interval gives the reader more information than the p-value alone. Because 0.79–1.22 includes values below and above 1, the reported data are compatible with a range of relative-hazard values under the confidence-interval framework.
19. Limitations and Interpretation Issues
- Registry-level detail: The ClinicalTrials.gov record contains endpoint definitions, statistical methods, estimates, confidence intervals, p-values, and selected safety counts, but not a complete statistical analysis plan.
- Hazard-ratio assumptions: Hazard ratios are time-to-event summaries whose interpretation depends on the underlying statistical framework. The ClinicalTrials.gov record does not permit an independent assessment of proportional hazards.
- Secondary endpoint multiplicity: Three secondary endpoint analyses are reported, but the ClinicalTrials.gov record does not specify a multiplicity-adjustment strategy for those analyses.
- No absolute endpoint estimates reported: The registry data gives the time frames for the 5-year and 8-year analyses but does not provide the corresponding arm-specific percentages in the posted statistical-analysis records.
- No reconstructed Kaplan-Meier curves: The ClinicalTrials.gov record does not contain participant-level event and censoring times needed to reconstruct survival curves.
- Safety comparison: Serious adverse-event counts are reported descriptively as affected participants over participants at risk, without a formal comparative analysis in the ClinicalTrials.gov record.
- Incomplete endpoint definition in registry-reported extract: The disease-free survival definition provided in the trial data ends with the registry text "or censored at da"; no additional wording is reported here, so the missing portion is not reconstructed.
- No additional design assumptions: The ClinicalTrials.gov record does not support conclusions about crossover, interim monitoring, Bayesian methods, imputation, or stratification.
20. Why This Trial Matters Statistically
TEXT is a useful teaching example because the ClinicalTrials.gov record contains a compact but rich set of classical randomized-trial methods: randomization, ITT analysis, time-to-event endpoints, log-rank testing, hazard ratios, confidence intervals, and p-values. The four posted analyses also demonstrate why a clinical-trial statistician should distinguish a primary endpoint from secondary outcomes rather than interpreting every p-value in isolation.
| Concept | How it appears in TEXT |
|---|---|
| Randomization | The trial uses randomized allocation with two parallel treatment arms. |
| Intention-to-treat | All four posted statistical analyses use the ITT population. |
| Time-to-event endpoints | Disease-free survival, breast cancer-free interval, distant recurrence-free interval, and overall survival are analyzed as time-to-event outcomes. |
| Log-rank test | Reported as the statistical method for all four posted analyses. |
| Hazard ratio | The reported effect measure for every posted endpoint analysis. |
| Confidence intervals | Each posted hazard-ratio analysis includes a two-sided 95% CI. |
| P-values | Each endpoint has a reported inferential p-value. |
| Superiority testing | The registry identifies the hypothesis type as superiority. |
| Endpoint hierarchy | One primary endpoint is distinguished from three secondary endpoints. |
| Safety description | Serious adverse events are reported by randomized arm as affected participants over participants at risk. |
21. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
22. Related Statistical Calculators
23. Sources
- ClinicalTrials.gov: NCT00066703 — TEXT.
- Linked publication: PubMed record for PMID 24881463.
- Linked publication: PubMed record for PMID 36521078.
- Linked publication: PubMed record for PMID 41637791.
- Linked publication: PubMed record for PMID 40986647.
- Linked publication: PubMed record for PMID 31618131.
Continue with the underlying statistical methods
Explore the survival-analysis, clinical-trial, and statistical-inference methods represented in the TEXT analysis.
24. Record Summary
TEXT provides a clear example of how a randomized phase 3 trial can use time-to-event methodology to answer several related but distinct clinical questions. The primary disease-free survival analysis reported an HR of 0.717 with a two-sided 95% CI of 0.602–0.855 and P = .0002. Secondary analyses reported HRs of 0.664 for breast cancer-free interval, 0.777 for distant recurrence-free interval, and 0.98 for overall survival.
The statistical lesson is broader than any single estimate. The trial illustrates why randomized treatment assignment and ITT analysis matter, why log-rank testing is appropriate for many time-to-event comparisons, why hazard ratios should be accompanied by confidence intervals, and why p-values should not be mistaken for measures of effect size. It also demonstrates the importance of distinguishing primary from secondary endpoints and superiority testing from equivalence or non-inferiority frameworks.