This page separates reported trial results from statistical interpretation. Numerical results are taken from the ClinicalTrials.gov record. The registry provides the official trial record; this page explains the statistical design and reported estimates without adding results from other sources.
1. Trial at a Glance
OlympiA is a randomized, parallel-group, triple-masked phase 3 trial evaluating olaparib versus placebo as adjuvant treatment in patients with germline BRCA-mutated high-risk HER2-negative primary breast cancer. The registry reports one primary time-to-event endpoint, Invasive Disease Free Survival (IDFS), with a stratified log-rank analysis and a hazard ratio estimated from a stratified Cox proportional-hazards model.
| Feature | OlympiA |
|---|---|
| Trial name | OlympiA |
| NCT ID | NCT02032823 |
| Phase | Phase 3 |
| Condition | Breast Cancer |
| Population | Patients with germline BRCA-mutated high-risk HER2-negative primary breast cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Treatment |
| Enrollment | 1837 |
| Interventions | Olaparib and placebo |
| Lead sponsor | AstraZeneca |
| Sponsor type | Industry |
| Status | Active, not recruiting |
| Start | 22 April 2014 |
| Primary completion | 27 March 2020 |
2. Clinical Question
The central statistical question is whether assignment to olaparib, compared with placebo, changes the time until the first occurrence of an invasive disease-free survival event in patients with germline BRCA-mutated high-risk HER2-negative primary breast cancer.
Population
Patients with germline BRCA-mutated high-risk HER2-negative primary breast cancer.
Intervention
Olaparib as the study intervention.
Comparator
Placebo.
Primary question
Does olaparib change Invasive Disease Free Survival relative to placebo?
The registered hypothesis type is superiority. This matters because the analysis is asking whether the time-to-event distribution differs between randomized treatment groups in a direction consistent with a lower event hazard for olaparib, rather than asking whether two treatments are sufficiently similar within a predefined non-inferiority margin.
3. Trial Design
Participants were allocated to olaparib or placebo.
The registry identifies the design model as parallel.
The registry identifies the masking as triple.
The registered primary purpose is treatment.
Olaparib
- Study intervention: olaparib.
- Compared against placebo in the randomized parallel design.
- Serious adverse events reported in the registry: 79 affected participants out of 911 at risk.
Placebo
- Comparator intervention: placebo.
- Compared against olaparib in the randomized parallel design.
- Serious adverse events reported in the registry: 79 affected participants out of 904 at risk.
The registry reports 1837 participants enrolled and two study arms. The serious-adverse-event denominators posted on ClinicalTrials.gov for the two arms are 911 and 904, respectively. These denominators should not be silently substituted for the overall enrollment figure when describing the trial.
4. Randomization, Stratification, and Analysis Population
The posted analyses use the Full Analysis Set (FAS). The statistical-analysis records explicitly identify intention-to-treat analysis as an associated concept. The registry also reports stratified analysis for each of the three posted time-to-event analyses.
| Feature | Registry-supported description |
|---|---|
| Analysis population | Full Analysis Set (FAS) |
| Associated analysis concept | Intention-to-treat analysis |
| Stratification | Chemotherapy type, hormone receptor status, and prior platinum therapy |
| Pooling strategy | Pre-specified pooling strategy |
| Stratified test | Stratified log-rank test |
| Effect model | Stratified Cox proportional-hazards model |
The same three factors are described as the stratification factors for the log-rank test and the stratified Cox model. This alignment is statistically useful: the treatment comparison and the model-based effect estimate are constructed around the same prespecified stratification structure rather than using unrelated adjustment schemes.
5. Endpoints
| Endpoint | Registry definition / time frame | Endpoint type |
|---|---|---|
| Invasive Disease Free Survival (IDFS) | An IDFS event is defined as the first occurrence of loco-regional or distant recurrence or new cancer or death from any cause. Time frame: from date of randomisation to data cut off: 27 March 2020 (approximately 5 years 11 months). | Time-to-event |
| Distant Disease Free Survival (DDFS) | From date of randomisation to data cut off: 27 March 2020 (approximately 5 years 11 months). | Time-to-event |
| Overall Survival (OS) | From date of randomisation to data cut off: 12 July 2021 (approximately 7 years 3 months). | Time-to-event |
The registry identifies one registered primary endpoint: IDFS. DDFS and OS are represented in the posted statistical analyses as secondary endpoints. The analysis records for all three endpoints use the Full Analysis Set and compare olaparib with placebo.
6. Statistical Methodology
Stratified log-rank test
The registry reports the log-rank test as the statistical method for IDFS, DDFS, and OS. For each analysis, the log-rank test is stratified by chemotherapy type, hormone receptor status, and prior platinum therapy using a pre-specified pooling strategy.
A log-rank test compares the observed pattern of events over follow-up between randomized groups. Unlike a simple comparison of proportions at a single time point, it uses the ordering of event and censoring times throughout the follow-up period. Stratification allows the comparison to account for the prespecified strata rather than treating all participants as if they came from a single homogeneous risk set.
Stratified Cox proportional-hazards model
The registry states that the treatment hazard ratio is based on a stratified Cox proportional-hazards model. The same stratification factors used in the stratified log-rank test are used in the Cox model.
The registry explicitly notes that a hazard ratio below 1 indicates a lower risk with olaparib compared with the placebo arm. The hazard ratio is a relative time-to-event measure; it is not an absolute risk difference and does not state what proportion of individual participants benefited.
Full Analysis Set and intention-to-treat reasoning
The posted analyses use the Full Analysis Set and identify intention-to-treat analysis as an associated concept. The statistical value of analyzing participants according to randomized assignment is that the treatment groups remain anchored to the allocation process that created the comparison. This helps preserve the causal interpretation reported by randomization rather than redefining groups according to what happened after assignment.
Stratification
Stratification is especially important here because the registry explicitly specifies three factors: chemotherapy type, hormone receptor status, and prior platinum therapy. In a stratified log-rank analysis, the treatment comparison is made while respecting these strata. In the corresponding stratified Cox model, the hazard ratio is estimated under the same stratification structure.
Censoring and follow-up
Time-to-event analyses require a way to handle participants who have not experienced the event by the time their available follow-up ends. In a standard survival-analysis framework, such observations are censored rather than treated as if the event had occurred. The registry's endpoint structure is explicitly time-to-event, making this distinction fundamental to interpretation.
7. Primary Result: Invasive Disease Free Survival
The posted primary analysis evaluates IDFS from the date of randomisation to the 27 March 2020 data cutoff, described in the registry as approximately 5 years 11 months. The analysis population is the Full Analysis Set. Olaparib is compared with placebo using a stratified log-rank test, with the hazard ratio estimated from a stratified Cox proportional-hazards model.
Invasive Disease Free Survival
Two-sided 99.5% CI: 0.409–0.816
P = 0.0000073
Analysis population: Full Analysis Set · Method: stratified log-rank test
| Primary endpoint | Olaparib vs placebo |
|---|---|
| Endpoint | Invasive Disease Free Survival (IDFS) |
| Time frame | From date of randomisation to data cut off: 27 March 2020 (approximately 5 years 11 months) |
| Analysis population | Full Analysis Set (FAS) |
| Method | Log Rank; stratified by chemotherapy type, hormone receptor status, and prior platinum therapy |
| Effect measure | Hazard Ratio |
| Estimate | 0.581 |
| Confidence interval | 99.5% two-sided CI: 0.409–0.816 |
| P-value | 0.0000073 |
| Hypothesis | Superiority |
What the estimate means: An HR of 0.581 means that the estimated instantaneous hazard of an IDFS event under the fitted stratified Cox model was about 58.1% of the corresponding hazard with placebo. Expressed as a simple relative-hazard interpretation, 0.581 corresponds to an estimated 41.9% lower hazard because 1 − 0.581 = 0.419.
What it does not mean: The HR does not mean that 58.1% of participants remained disease-free, that 41.9% of participants avoided an event, or that every individual participant experienced exactly a 41.9% reduction in risk. It is a model-based relative comparison of event hazards.
What the confidence interval says: The two-sided 99.5% confidence interval extends from 0.409 to 0.816. It quantifies uncertainty around the estimated hazard ratio under the statistical model and sampling framework. It does not describe the range of effects experienced by individual patients.
Why the p-value is different from the effect size: The P-value of 0.0000073 addresses the statistical evidence against the null comparison under the reported testing framework. It is not a measure of how large the treatment effect is. The magnitude of the estimated effect is conveyed by the HR, while the confidence interval describes its precision.
Important cautions: The analysis is time-to-event based and uses a stratified Cox proportional-hazards model. Interpretation of a single HR therefore depends on the model framework, including the proportional-hazards interpretation. Censoring and follow-up also matter because IDFS is not a simple binary endpoint at one fixed time point. The analysis population is the Full Analysis Set, and the registry identifies the test as a superiority analysis.
8. Secondary Result: Distant Disease Free Survival
Distant Disease Free Survival is reported as a secondary time-to-event endpoint. The registry analysis uses the Full Analysis Set and evaluates olaparib versus placebo from randomisation to the 27 March 2020 data cutoff, approximately 5 years 11 months.
Distant Disease Free Survival
Two-sided 99.5% CI: 0.392–0.831
P = 0.0000257
Analysis population: Full Analysis Set · Method: stratified log-rank test
| Secondary endpoint | Olaparib vs placebo |
|---|---|
| Endpoint | Distant Disease Free Survival (DDFS) |
| Time frame | From date of randomisation to data cut off: 27 March 2020 (approximately 5 years 11 months) |
| Analysis population | Full Analysis Set (FAS) |
| Method | Log Rank; stratified by chemotherapy type, hormone receptor status, and prior platinum therapy |
| Effect measure | Hazard Ratio |
| Estimate | 0.574 |
| Confidence interval | 99.5% two-sided CI: 0.392–0.831 |
| P-value | 0.0000257 |
| Hypothesis | Superiority |
What the estimate means: The DDFS HR of 0.574 corresponds to an estimated instantaneous event hazard about 57.4% as high with olaparib as with placebo under the reported stratified Cox model. As a direct arithmetic interpretation, 1 − 0.574 = 0.426, or an estimated 42.6% lower hazard.
What it does not mean: It does not mean that 42.6% of participants were prevented from developing a distant event, nor does it give an individual participant's probability of remaining distant-disease-free.
Precision: The two-sided 99.5% confidence interval is 0.392–0.831. The interval communicates uncertainty around the HR estimate; it should be read separately from the P-value.
Testing versus estimation: The P-value of 0.0000257 addresses the reported statistical comparison. It should not be used as a substitute for examining the magnitude and precision of the hazard ratio.
Analysis cautions: DDFS is a time-to-event endpoint, so censoring and the pattern of follow-up matter. The analysis is stratified by the same three prespecified factors as the IDFS analysis, and the proportional-hazards model supplies the reported HR.
9. Secondary Result: Overall Survival
Overall Survival is another secondary time-to-event endpoint. Its posted analysis uses a later data cutoff: 12 July 2021, approximately 7 years 3 months from randomisation. As with IDFS and DDFS, the Full Analysis Set is used and the treatment comparison is performed with a stratified log-rank test.
Overall Survival
Two-sided 98.5% CI: 0.468–0.973
P = 0.0091
Analysis population: Full Analysis Set · Method: stratified log-rank test
| Secondary endpoint | Olaparib vs placebo |
|---|---|
| Endpoint | Overall Survival (OS) |
| Time frame | From date of randomisation to data cut off: 12 July 2021 (approximately 7 years 3 months) |
| Analysis population | Full Analysis Set (FAS) |
| Method | Log Rank; stratified by chemotherapy type, hormone receptor status, and prior platinum therapy |
| Effect measure | Hazard Ratio |
| Estimate | 0.678 |
| Confidence interval | 98.5% two-sided CI: 0.468–0.973 |
| P-value | 0.0091 |
| Hypothesis | Superiority |
What the estimate means: An OS HR of 0.678 means that the estimated instantaneous hazard of death under the reported stratified Cox model was about 67.8% of the corresponding hazard with placebo. Numerically, 1 − 0.678 = 0.322, corresponding to an estimated 32.2% lower hazard.
What it does not mean: The HR does not mean that 32.2% of patients survived, that 32.2% of patients were cured, or that each participant had exactly a 32.2% reduction in mortality risk.
Precision: The two-sided 98.5% confidence interval is 0.468–0.973. The interval is the appropriate place to examine the uncertainty around the estimated HR; the P-value alone cannot communicate that precision.
P-value versus effect size: The P-value of 0.0091 describes the evidence under the reported statistical testing framework. It does not quantify the size or clinical importance of the effect. The HR and its confidence interval are the principal effect-estimation quantities reported here.
Different data cutoff: The OS analysis uses 12 July 2021, whereas IDFS and DDFS use 27 March 2020. These estimates therefore should not be treated as though they came from the same data cutoff simply because all three are hazard ratios.
10. Results in One Statistical View
| Endpoint | Role | Data cutoff | HR | CI | P-value |
|---|---|---|---|---|---|
| Invasive Disease Free Survival | Primary | 27 March 2020 | 0.581 | 99.5% CI 0.409–0.816 | 0.0000073 |
| Distant Disease Free Survival | Secondary | 27 March 2020 | 0.574 | 99.5% CI 0.392–0.831 | 0.0000257 |
| Overall Survival | Secondary | 12 July 2021 | 0.678 | 98.5% CI 0.468–0.973 | 0.0091 |
The three estimates are directionally similar in the registry analyses: each HR is below 1. The most important statistical qualification is that the endpoints have different roles and, for OS, a different data cutoff and confidence level. A single table can therefore summarize them, but it should not erase those distinctions.
The visual above is deliberately a display of the reported HR estimates rather than a plot of survival probability. A hazard ratio is not a percentage of patients and should not be interpreted as one.
11. Safety Results
The ClinicalTrials.gov record reports serious adverse events by randomized arm. These data provide an affected-participant count and an at-risk denominator for each arm.
| Safety measure | Olaparib | Placebo |
|---|---|---|
| Serious adverse events, affected / at risk | 79 / 911 | 79 / 904 |
Olaparib arm
The registry reports 79 affected participants among 911 at risk for serious adverse events.
Placebo arm
The registry reports 79 affected participants among 904 at risk for serious adverse events.
The equal affected-participant count does not mean that the two arms have identical serious-adverse-event rates, because the registry-reported denominators differ. The appropriate comparison of these registry figures is therefore based on the reported affected and at-risk counts rather than on the raw counts alone.
12. Statistical Methods Explained
Why was a stratified log-rank test used?
The registered endpoints are time-to-event outcomes, and the reported method is a log-rank test. Stratification incorporates the prespecified factors of chemotherapy type, hormone receptor status, and prior platinum therapy into the comparison. This preserves the trial's planned stratified analysis rather than replacing it with an unstratified comparison.
What does an IDFS hazard ratio of 0.581 mean?
Under the reported stratified Cox model, the estimated instantaneous IDFS-event hazard with olaparib is 0.581 times that with placebo. The direct arithmetic interpretation is a 41.9% lower estimated hazard because 1 − 0.581 = 0.419. This is not a statement that 41.9% of participants were protected from an event.
Why is the confidence interval so important?
The point estimate alone does not communicate statistical precision. For IDFS, the two-sided 99.5% confidence interval is 0.409–0.816. For DDFS it is 0.392–0.831, and for OS it is 0.468–0.973 with a 98.5% confidence level. These intervals show the uncertainty associated with the corresponding estimates.
Why does the P-value not measure effect size?
A P-value addresses the statistical evidence against a null hypothesis under the specified analysis. It does not tell the reader whether the observed effect is large, small, clinically important, or practically meaningful. Those questions require examination of the effect estimate, confidence interval, endpoint definition, and clinical context.
Why does the Full Analysis Set matter?
The posted analyses identify the Full Analysis Set as the analysis population and list intention-to-treat analysis among the associated concepts. Anchoring efficacy analysis to randomized assignment helps preserve the treatment comparison established by randomization rather than conditioning the comparison on events that occur after assignment.
What does stratification accomplish in the Cox model?
The registry states that the hazard ratio is estimated from a stratified Cox proportional-hazards model using the same stratification factors as the stratified log-rank test. Stratification allows those factors to define the analysis structure while estimating the treatment hazard ratio across the strata.
What is the proportional-hazards caution?
A Cox hazard ratio is a model-based summary of relative event hazards. Its most straightforward interpretation assumes that the relative hazards represented by the model are appropriately summarized by a single HR over the analyzed period. If that assumption is not a good description of the underlying event processes, a single HR may compress a more complex time-varying treatment effect.
13. Confidence Intervals and Statistical Evidence
The OlympiA registry provides unusually informative confidence levels for the posted analyses: IDFS and DDFS use two-sided 99.5% confidence intervals, while OS uses a two-sided 98.5% confidence interval. The confidence level should always be read alongside the estimate rather than omitted from the result.
| Endpoint | Estimate | Confidence level | Interval |
|---|---|---|---|
| IDFS | 0.581 | 99.5% | 0.409–0.816 |
| DDFS | 0.574 | 99.5% | 0.392–0.831 |
| OS | 0.678 | 98.5% | 0.468–0.973 |
The interval is an uncertainty statement about the estimated treatment effect under the specified statistical framework. It is not a prediction interval for an individual participant and does not mean that individual treatment effects are distributed uniformly between the lower and upper limits.
The fact that all three reported intervals lie below 1 is consistent with the direction of the corresponding HR estimates. The P-values provide a separate summary of statistical evidence and should not be used to replace the interval-based assessment of precision.
14. The Importance of the Different Data Cutoffs
Trial start
The registry profile lists 22 April 2014 as the trial start date.
Primary completion and IDFS / DDFS cutoff
The registry lists 27 March 2020 as the primary completion date. The IDFS and DDFS analyses use this date as their data cutoff and describe the follow-up as approximately 5 years 11 months.
OS data cutoff
The posted OS analysis uses 12 July 2021 as its data cutoff, described as approximately 7 years 3 months from randomisation.
This distinction is more than a documentation detail. Time-to-event estimates depend on how much follow-up is available. The IDFS and DDFS estimates and the OS estimate therefore summarize different observation windows, even though all three analyses use the Full Analysis Set.
15. Primary Analysis vs Secondary Analyses
| Statistical feature | IDFS | DDFS | OS |
|---|---|---|---|
| Endpoint role | Primary | Secondary | Secondary |
| Endpoint type | Time-to-event | Time-to-event | Time-to-event |
| Analysis population | Full Analysis Set | Full Analysis Set | Full Analysis Set |
| Method | Log-rank | Log-rank | Log-rank |
| Stratified analysis | Yes | Yes | Yes |
| Effect measure | Hazard ratio | Hazard ratio | Hazard ratio |
| Hypothesis type | Superiority | Superiority | Superiority |
This structure illustrates why a clinical-trial results page should not collapse every reported P-value into one undifferentiated list. The primary endpoint establishes the principal registered efficacy question, while the secondary endpoints address additional outcomes using the same general survival-analysis framework.
16. What the Hazard Ratios Do — and Do Not — Mean
The HR of 0.581 indicates a lower estimated IDFS-event hazard with olaparib under the reported stratified Cox model. Numerically, it corresponds to a 41.9% lower estimated hazard relative to placebo.
It does not indicate that 41.9% of participants avoided recurrence or death, and it does not describe an individual patient's probability of remaining event-free.
The HR of 0.574 indicates a lower estimated DDFS-event hazard with olaparib under the reported model, corresponding arithmetically to a 42.6% lower estimated hazard.
It is not equivalent to an absolute 42.6 percentage-point difference in distant disease-free survival.
The HR of 0.678 indicates a lower estimated mortality hazard with olaparib under the reported model, corresponding arithmetically to a 32.2% lower estimated hazard.
It does not mean that 32.2% of patients survived or that each patient experienced a 32.2% reduction in mortality risk.
The common feature is that each HR is a relative time-to-event measure. The numerical values should therefore be interpreted with the endpoint definition, follow-up period, confidence interval, analysis population, and model specification kept visible.
17. Limitations
- Registry-level detail: The ClinicalTrials.gov record provides the registered endpoint definitions and posted statistical analyses but do not provide a complete statistical analysis plan. Conclusions about unreported design features should therefore not be inferred.
- Different data cutoffs: IDFS and DDFS use 27 March 2020, whereas OS uses 12 July 2021. The estimates describe different follow-up windows.
- Hazard-ratio interpretation: A single Cox HR is a model-based relative measure. Its interpretation depends on the proportional-hazards framework represented by the model.
- Censoring: Time-to-event analysis necessarily involves participants whose event status is not observed through the entire potential follow-up period. The validity of survival estimates depends on appropriate handling and assumptions about censoring.
- Secondary endpoints: DDFS and OS are secondary endpoints in the ClinicalTrials.gov record. Their statistical role is therefore different from that of the registered primary IDFS endpoint.
- Confidence levels differ: The analyses posted on ClinicalTrials.gov use 99.5% confidence intervals for IDFS and DDFS and a 98.5% confidence interval for OS. These should not be silently converted to a common confidence level.
- Limited safety information: The ClinicalTrials.gov record contains serious adverse-event counts and denominators by arm but do not provide a full safety profile.
- No unsupported reconstruction: The ClinicalTrials.gov record does not contain Kaplan-Meier coordinates, median event times, subgroup estimates, or individual event counts. Those quantities are therefore not reconstructed here.
18. Why This Trial Matters Statistically
OlympiA is a useful teaching case because the ClinicalTrials.gov record connects a randomized phase 3 design with a primary time-to-event endpoint, a stratified log-rank comparison, a stratified Cox hazard ratio, an intention-to-treat concept, and multiple follow-up cutoffs. Those elements illustrate how a modern clinical-trial result is built from several related statistical decisions rather than from a P-value alone.
| Concept | How it appears in OlympiA |
|---|---|
| Randomization | The trial is randomized with two parallel arms. |
| Blinding | The registry identifies the trial as triple masked. |
| Time-to-event analysis | IDFS is the registered primary endpoint; DDFS and OS are posted secondary endpoints. |
| Log-rank test | Reported as the method for all three posted statistical analyses. |
| Stratified analysis | Analyses are stratified by chemotherapy type, hormone receptor status, and prior platinum therapy. |
| Hazard ratio | Used as the effect measure for IDFS, DDFS, and OS. |
| Cox model | The treatment HR is based on a stratified Cox proportional-hazards model. |
| Confidence intervals | 99.5% intervals are reported for IDFS and DDFS; 98.5% for OS. |
| Intention-to-treat principle | Listed among the concepts associated with the posted analyses; the analysis population is the Full Analysis Set. |
| Endpoint hierarchy | IDFS is primary; DDFS and OS are secondary. |
| Follow-up maturity | IDFS/DDFS and OS use different data cutoffs and therefore different follow-up windows. |
| Safety comparison | Serious adverse events are reported as affected participants divided by participants at risk in each arm. |
The statistical lesson is that the treatment estimate cannot be separated from the endpoint definition and analysis framework. A hazard ratio without its endpoint, data cutoff, confidence interval, analysis population, and model would be an incomplete description of the evidence.
19. Statistical Methods: A Practical Walk-Through
Step 1: Start with the randomized comparison
The trial has two interventions, olaparib and placebo, within a randomized parallel design. Randomization establishes the basic comparison that the efficacy analysis is intended to preserve.
Step 2: Define the time-to-event endpoint
For the primary endpoint, IDFS begins at randomisation and ends at the first occurrence of a registry-defined IDFS event or the appropriate censoring point. The registry-reported definition is specific: the first occurrence of loco-regional or distant recurrence, new cancer, or death from any cause.
Step 3: Compare event patterns with a stratified log-rank test
The log-rank framework compares the observed event experience between randomized groups across follow-up. The analysis is stratified by chemotherapy type, hormone receptor status, and prior platinum therapy according to the registry analysis notes.
Step 4: Estimate the relative effect
The treatment hazard ratio is obtained from a stratified Cox proportional-hazards model using the same stratification factors. For IDFS, the resulting HR is 0.581; for DDFS it is 0.574; and for OS it is 0.678.
Step 5: Quantify uncertainty
Each estimate is paired with a confidence interval. The confidence level is not identical across all three analyses, so the interval should be reported exactly as posted rather than converted to another level.
Step 6: Interpret the P-value separately
The P-value is evidence under the statistical testing framework. It should not be substituted for the HR or its confidence interval. The magnitude and precision of the effect are different questions from the strength of evidence against the null hypothesis.
20. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The posted analyses report hazard ratios below 1 for IDFS, DDFS, and OS, with corresponding confidence intervals and P-values. The primary IDFS analysis is a superiority analysis based on a stratified log-rank test with the HR estimated from a stratified Cox model.
Clinical interpretation
The clinical meaning of these estimates depends on the definition of each endpoint, the duration of follow-up, the magnitude and precision of the treatment effect, and the broader clinical context. The ClinicalTrials.gov record alone do not provide enough information to add unsupported clinical outcomes or treatment recommendations.
This distinction is important. Statistical evidence describes what the analysis estimates and how uncertain that estimate is. Clinical interpretation asks what that evidence means in the context of patient outcomes. A rigorous trial-results page should keep those layers connected but distinct.
21. A Closer Look at the Primary IDFS Analysis
The primary analysis combines a nonparametric-style comparison of event experience over time with a model-based estimate of relative hazard. Both components use the registry-specified stratification factors.
The primary IDFS result is therefore more informative than the P-value alone. The complete statistical statement is the endpoint, analysis population, data cutoff, test, stratification factors, HR, confidence interval, and P-value taken together.
| Component | Primary IDFS analysis |
|---|---|
| Endpoint | Invasive Disease Free Survival |
| Definition | First occurrence of loco-regional or distant recurrence or new cancer or death from any cause |
| Time frame | From date of randomisation to data cut off: 27 March 2020 (approximately 5 years 11 months) |
| Population | Full Analysis Set |
| Comparison | Olaparib vs placebo |
| Test | Stratified log-rank |
| Strata | Chemotherapy type; hormone receptor status; prior platinum therapy |
| Model for HR | Stratified Cox proportional-hazards model |
| HR | 0.581 |
| 99.5% CI | 0.409–0.816 |
| P-value | 0.0000073 |
22. What the Registry Data Do Not Establish
A disciplined statistical analysis also requires knowing where the registry-reported evidence stops. The trial data identify three posted statistical analyses and provide their estimates, confidence intervals, P-values, populations, methods, and data cutoffs. They do not provide every possible quantity that a full clinical-trial publication might contain.
No median event times reported
The ClinicalTrials.gov record does not provide median IDFS, DDFS, or OS values. They are therefore not reported here.
No subgroup estimates reported
The ClinicalTrials.gov record contains no subgroup hazard ratios or interaction tests.
No KM coordinates reported
The available data do not contain the event and censoring information required to reconstruct a Kaplan-Meier curve.
No interim rules reported
The ClinicalTrials.gov record does not state an interim-analysis schedule, alpha-spending strategy, or stopping boundary.
This is not a limitation of survival analysis itself. It is a limitation of the specific information available in the ClinicalTrials.gov record. Adding values from memory or from an unlisted publication would violate the data boundary for this page.
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Calculators
The most directly relevant calculator topics follow the trial's time-to-event and inference pathway:
25. Sources
- ClinicalTrials.gov: OlympiA — NCT02032823.
- PubMed: PMID 38301187.
- PubMed: PMID 35594464.
- PubMed: PMID 34081848.
Continue with the underlying statistical methods
Use the trial's endpoints and analysis methods as a practical route into survival analysis, hazard ratios, confidence intervals, randomization, and related clinical-trial methods.
26. Record Summary
OlympiA provides a clear example of a randomized phase 3 time-to-event analysis. The ClinicalTrials.gov record identifies one primary endpoint, Invasive Disease Free Survival, and two posted secondary analyses, Distant Disease Free Survival and Overall Survival. All three use the Full Analysis Set, a stratified log-rank test, and a hazard ratio estimated from a stratified Cox proportional-hazards model. The stratification factors are chemotherapy type, hormone receptor status, and prior platinum therapy.
The primary IDFS analysis reports an HR of 0.581 with a two-sided 99.5% CI of 0.409–0.816 and P = 0.0000073. The secondary DDFS analysis reports an HR of 0.574 with a two-sided 99.5% CI of 0.392–0.831 and P = 0.0000257. The secondary OS analysis, using a later data cutoff, reports an HR of 0.678 with a two-sided 98.5% CI of 0.468–0.973 and P = 0.0091.
The most useful statistical reading of these results is not simply that all three P-values are small. The more complete interpretation combines the endpoint definition, randomized comparison, analysis population, stratification, log-rank test, Cox model, hazard ratio, confidence interval, P-value, and data cutoff. That framework makes clear both what the reported estimates show and what they do not establish.