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
RE-SONATE was a randomized, double-blind, parallel-group phase 3 trial evaluating dabigatran etexilate 150 mg twice daily versus matching placebo for prevention of recurrent symptomatic venous thromboembolism during an intended 6-month treatment period.
| Feature | RE-SONATE |
|---|---|
| Phase | Phase 3 |
| Condition | Venous Thromboembolism |
| Purpose | Prevention |
| Design | Randomized, double-blind, parallel |
| Allocation | Randomized |
| Enrollment | 1,353 |
| Intervention | Dabigatran etexilate 150 mg twice daily (BID) |
| Comparator | Matching placebo twice daily (BID) |
| Primary endpoint | Centrally confirmed symptomatic recurrent VTE including unexplained death during the intended treatment period |
| Primary endpoint time frame | 6 months |
| Primary endpoint type | Time-to-event |
| Primary hypothesis | Superiority |
| Results posted | Yes |
| Statistical analyses posted | 16 |
2. Clinical Question
The central statistical question was whether dabigatran etexilate 150 mg twice daily reduced the time-to-event risk of centrally confirmed symptomatic recurrent venous thromboembolic events, including unexplained death, compared with matching placebo during the intended 6-month treatment period.
Population
Participants enrolled in the phase 3 prevention trial for venous thromboembolism.
Intervention
Dabigatran etexilate 150 mg twice daily.
Comparator
Matching placebo twice daily.
Primary question
Does dabigatran reduce the hazard of centrally confirmed symptomatic recurrent VTE including unexplained death during the intended treatment period?
3. Trial Design
Dabigatran etexilate
- 150 mg twice daily (BID)
- Double-blind treatment
- Intended treatment period of 6 months
- Primary efficacy analysis against placebo
Matching placebo
- Twice daily (BID)
- Double-blind treatment
- Intended treatment period of 6 months
- Primary efficacy analysis against dabigatran
The registry describes a randomized, double-blind, parallel design. Randomization provides the principal basis for comparing the treatment groups, while double blinding is intended to reduce the influence of treatment knowledge on participant behavior, clinical management, assessment, and reporting.
4. Endpoints
| Role | Endpoint | Time frame | Type |
|---|---|---|---|
| Primary | Centrally Confirmed Symptomatic Recurrent Venous Thrombotic Events (VTE) Including Unexplained Death During the Intended Treatment Period | 6 months | Time-to-event |
| Secondary | Centrally Confirmed Symptomatic Recurrent Venous Thrombotic Events (VTE) Excluding Unexplained Death During the Intended Treatment Period | 6 months | Time-to-event |
| Secondary | Centrally Confirmed Symptomatic Recurrent Deep Venous Thrombotic (DVT) Events During the Intended Treatment Period | 6 months | Binary |
| Secondary | Centrally Confirmed Symptomatic Pulmonary Embolism (PE) Events During the Intended Treatment Period | 6 months | Binary |
| Secondary | Centrally Confirmed Unexplained Deaths During the Intended Treatment Period | 6 months | Binary |
| Secondary | Centrally Confirmed Bleeding Event During the Treatment Period | 6 months | Binary and time-to-event analyses |
The primary endpoint is a composite time-to-event outcome. The registry defines symptomatic recurrent VTE as the composite of recurrent deep vein thrombosis, fatal or non-fatal pulmonary embolism, and unexplained death within the specified endpoint definition. Although the registry reports participant event counts for the outcome, its primary effect measure is a hazard ratio representing the time-to-event comparison.
5. Analysis Populations
| Population | Definition / role |
|---|---|
| Full analysis set | Randomized and treated participants. |
| Primary efficacy analysis | Full analysis set, analyzed as randomized. |
| Bleeding analyses | Full analysis set, analyzed as treated. |
The distinction between “as randomized” and “as treated” is statistically important. The primary VTE analysis retains randomized treatment assignment, which preserves the comparison created by randomization. The bleeding analyses were conducted according to treatment received, which is more directly aligned with observed treatment exposure but does not have exactly the same causal interpretation as an analysis anchored to randomized assignment.
6. Statistical Methodology
Cox proportional-hazards model
The primary endpoint was analyzed using a Cox proportional-hazards model. This is appropriate for a time-to-event endpoint because it uses the ordering and timing of events while allowing participants who do not experience the event during observation to contribute censored follow-up.
An HR below 1 indicates a lower estimated instantaneous event rate in the dabigatran group under the fitted model. The HR is not an absolute risk difference and is not the percentage of participants who experienced an event.
Exact / Clopper-Pearson binomial confidence intervals
Several secondary binary endpoints were summarized as percentages of participants with events, with two-sided 95% confidence intervals calculated using the exact Clopper-Pearson method. This method is especially useful when event counts are small because it does not rely on the normal approximation to the binomial distribution.
Fisher exact test
The registry also reports Fisher exact tests for comparisons of DVT, pulmonary embolism, unexplained death, and bleeding outcomes. Fisher's exact test evaluates the treatment-group association for a binary outcome using the exact distribution conditional on the observed margins.
Time-to-event versus binary analysis
The same clinical outcome can support different statistical summaries depending on the analysis question. A binary analysis asks whether an event occurred during a defined period. A time-to-event analysis additionally uses when the event occurred and accounts for censoring. The primary RE-SONATE analysis used the latter framework.
Two participants with the same final event status can contribute different amounts of information if one is followed for substantially longer before an event or censoring. Cox regression incorporates this timing rather than reducing the entire follow-up period to a single yes/no indicator.
7. Primary Result: Recurrent Symptomatic VTE Including Unexplained Death
The primary endpoint was centrally confirmed symptomatic recurrent VTE including unexplained death during the intended treatment period, assessed over 6 months. The full analysis set was analyzed as randomized using a Cox proportional-hazards model.
Hazard ratio for the primary endpoint
95% CI: 0.02–0.25 · P < 0.0001
Dabigatran vs placebo · Superiority analysis
An HR of 0.08 means that, under the fitted Cox model, the estimated instantaneous hazard of the primary VTE endpoint in the dabigatran group was approximately 92% lower than the estimated hazard in the placebo group during the analyzed treatment period.
The HR does not mean that 92% of participants were protected, that the absolute probability of recurrence was reduced by 92%, or that every participant experienced the same proportional reduction. It is a relative time-to-event measure describing the treatment groups under the model.
The 95% confidence interval of 0.02–0.25 describes uncertainty around the estimated hazard ratio. It remains below 1 throughout the reported interval, indicating that the estimated treatment effect is consistently in the direction of lower hazard across the interval of values supported by this analysis.
The P < 0.0001 value addresses the statistical evidence against the null hypothesis associated with the reported superiority analysis. It does not measure the magnitude of the treatment effect, and it should not be interpreted as the probability that the treatment is effective.
Because this is a Cox analysis, interpretation also depends on the proportional-hazards framework. The registry does not provide a separate assessment of the proportional-hazards assumption in the information reported here.
What the primary result establishes statistically
The reported analysis provides a direct randomized comparison of time to centrally confirmed symptomatic recurrent VTE including unexplained death. The effect estimate, confidence interval, and P-value all describe the same prespecified superiority comparison; they answer different questions and should be interpreted together rather than treating the P-value as a substitute for the effect estimate.
8. Secondary Time-to-Event Endpoint: Recurrent VTE Excluding Unexplained Death
The registry reports a secondary analysis of centrally confirmed symptomatic recurrent VTE excluding unexplained death during the intended treatment period. This was analyzed over 6 months in the full analysis set and analyzed as randomized using a Cox proportional-hazards model.
Hazard ratio for recurrent VTE excluding unexplained death
95% CI: 0.03–0.27 · P < 0.0001
Dabigatran vs placebo · Superiority analysis
An HR of 0.08 corresponds to an estimated instantaneous event hazard approximately 92% lower with dabigatran than placebo under the fitted Cox model for this secondary endpoint.
The 95% CI of 0.03–0.27 quantifies precision around the hazard-ratio estimate. It does not describe individual-level treatment response or provide an absolute percentage reduction in VTE.
The reported P < 0.0001 provides evidence against the null hypothesis for this superiority comparison. As with the primary analysis, the P-value is not a measure of effect size.
This endpoint differs from the primary composite because unexplained death is excluded. Consequently, its result should be interpreted as a distinct endpoint rather than as a duplicate numerical statement of the primary analysis.
9. Secondary Binary VTE Components
Deep Venous Thrombotic Events
The registry reports centrally confirmed symptomatic recurrent DVT during the intended treatment period as a secondary binary endpoint. The full analysis set was analyzed as randomized.
| Group | Percentage with DVT | 95% Clopper-Pearson CI |
|---|---|---|
| Dabigatran | 0.3 | 0.04–1.06 |
| Placebo | 3.5 | 2.21–5.17 |
The registry also reports a Fisher exact test comparing dabigatran with placebo, with P < 0.0001.
The reported percentages describe the proportion of participants experiencing the binary DVT endpoint during the specified period. The exact confidence intervals reflect uncertainty around each group's event percentage and are calculated separately for each group.
The Fisher exact P-value addresses the treatment-group comparison, whereas the Clopper-Pearson intervals address the precision of the individual group percentages. Neither the confidence interval nor the P-value should be interpreted as a direct estimate of an individual's probability of benefit.
Pulmonary Embolism
Centrally confirmed symptomatic PE during the intended treatment period was also evaluated as a secondary binary endpoint.
| Group | Percentage with PE | 95% Clopper-Pearson CI |
|---|---|---|
| Dabigatran | 0.1 | 0.00–0.82 |
| Placebo | 2.1 | 1.16–3.52 |
The Fisher exact test comparing the groups produced P = 0.0004.
The estimated PE percentages are small in both groups, with the dabigatran confidence interval extending from 0.00 to 0.82. That interval illustrates why small observed event percentages can still have meaningful statistical uncertainty.
The Fisher exact P-value evaluates the observed treatment-group difference under the exact test framework. It does not quantify the magnitude of the difference, while the group-specific confidence intervals do not themselves constitute a formal confidence interval for the between-group contrast.
Unexplained Death
Centrally confirmed unexplained deaths during the intended treatment period were evaluated as another secondary binary endpoint.
| Group | Percentage with unexplained death | 95% Clopper-Pearson CI |
|---|---|---|
| Dabigatran | 0.0 | 0.00–0.54 |
| Placebo | 0.3 | 0.04–1.09 |
The Fisher exact comparison produced P = 0.2428.
An observed percentage of 0.0 does not mean the underlying event probability is exactly zero. The reported 95% Clopper-Pearson interval for dabigatran extends from 0.00 to 0.54, demonstrating that a zero observed event percentage can still be compatible with a range of underlying probabilities.
The reported Fisher exact P-value of 0.2428 does not provide evidence against the null hypothesis at conventional significance levels. It should not be converted into a statement that the treatments are equivalent or that there is no possible difference.
10. Secondary Bleeding Results
Bleeding was evaluated separately from the VTE efficacy endpoints. The registry reports both binary percentages with exact confidence intervals and time-to-first-event analyses using Cox proportional-hazards models.
Bleeding Event: Binary Analysis
| Group | Percentage with bleeding event | 95% Clopper-Pearson CI |
|---|---|---|
| Dabigatran | 0.3 | 0.04–1.05 |
| Placebo | 0.0 | 0.00–0.56 |
The registry reports the binary bleeding comparison using Fisher exact testing, with P = 0.4998. The analysis was conducted in the full analysis set as treated; 3 participants randomized to placebo were treated with dabigatran only.
The binary analysis describes the proportion of participants with the reported bleeding endpoint. The confidence intervals are exact two-sided 95% intervals calculated separately within each treatment group.
The Fisher exact P-value of 0.4998 does not provide evidence of a statistically detectable treatment-group difference under this particular binary comparison. It does not establish equality of the two treatment groups or rule out clinically meaningful differences that the available event information may not estimate precisely.
Time to First Major or Clinically Relevant Bleeding Event
Hazard ratio for time to first occurrence of an MBE or CRBE
95% CI: 1.52–5.60 · P = 0.0013
Dabigatran vs placebo · FAS, as treated
An HR of 2.92 means that the estimated instantaneous hazard of the first occurrence of the reported MBE or CRBE endpoint was approximately 2.92 times as high in the dabigatran group as in the placebo group under the fitted Cox model.
This is the opposite direction from the VTE efficacy hazard ratio. An HR above 1 does not mean that 2.92 times as many participants necessarily experienced the event; it is a relative time-to-event measure.
The 95% CI of 1.52–5.60 describes uncertainty around the hazard-ratio estimate. The interval lies above 1, while the reported P-value of 0.0013 provides statistical evidence against the corresponding null hypothesis.
The as-treated analysis is particularly important here because treatment received, rather than randomized assignment, determines the analysis group. The registry identifies 3 participants who were randomized to placebo but treated with dabigatran.
Time to First Occurrence of Any Bleeding Event
Hazard ratio for time to first bleeding event
95% CI: 1.23–2.68 · P = 0.0027
Dabigatran vs placebo · FAS, as treated
An HR of 1.82 corresponds to an estimated instantaneous hazard of first bleeding approximately 82% higher in the dabigatran group than in the placebo group under the fitted Cox model.
The 95% CI of 1.23–2.68 indicates uncertainty around the estimated relative hazard while remaining above 1. The P-value of 0.0027 addresses the statistical evidence for the group comparison rather than measuring the size or clinical importance of the bleeding difference.
Because this is a time-to-event endpoint, the analysis uses the timing of first bleeding and censoring rather than simply classifying every participant as having or not having a bleeding event by 6 months.
11. Summary of Reported Statistical Analyses
| Endpoint | Analysis | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Primary recurrent VTE including unexplained death | Cox proportional-hazards model | Hazard ratio | 0.08 | 0.02–0.25 | <0.0001 |
| Recurrent VTE excluding unexplained death | Cox proportional-hazards model | Hazard ratio | 0.08 | 0.03–0.27 | <0.0001 |
| Bleeding: time to MBE or CRBE | Cox proportional-hazards model | Hazard ratio | 2.92 | 1.52–5.60 | 0.0013 |
| Bleeding: time to any bleeding event | Cox proportional-hazards model | Hazard ratio | 1.82 | 1.23–2.68 | 0.0027 |
| DVT | Fisher exact test | P-value | — | — | <0.0001 |
| PE | Fisher exact test | P-value | — | — | 0.0004 |
| Unexplained death | Fisher exact test | P-value | — | — | 0.2428 |
| Bleeding event | Fisher exact test | P-value | — | — | 0.4998 |
The registry therefore contains two distinct statistical layers: time-to-event modeling for the primary endpoint and selected bleeding outcomes, and categorical analyses for several component or safety endpoints. Keeping these estimands separate prevents a binary event percentage from being confused with a hazard ratio.
12. Serious Adverse Events
The registry reports serious adverse events by treatment arm as affected participants divided by participants at risk.
| Arm | Affected / at risk |
|---|---|
| Dabigatran | 47 / 684 |
| Placebo | 60 / 659 |
These counts provide a direct descriptive safety summary. They should not be substituted for the separately analyzed bleeding endpoints, because serious adverse events and centrally confirmed bleeding events represent different outcome definitions.
Dabigatran
47 of 684 participants were affected by serious adverse events according to the registry summary.
Placebo
60 of 659 participants were affected by serious adverse events according to the registry summary.
13. Statistical Methods Explained
Why was a Cox proportional-hazards model used?
The primary endpoint is a time-to-event outcome observed during a defined treatment period. Cox regression allows the analysis to incorporate both whether an event occurred and when it occurred, while accounting for censored observations. The resulting hazard ratio summarizes the relative instantaneous event hazard between the treatment groups under the fitted model.
What does an HR of 0.08 mean?
An HR of 0.08 means the estimated instantaneous event hazard in the dabigatran group was 0.08 times the estimated hazard in the placebo group. Expressed as a relative reduction in the estimated hazard, this corresponds to approximately 92% lower estimated hazard. It does not mean a 92 percentage-point reduction in event probability.
Why is the confidence interval important?
The point estimate alone does not communicate statistical precision. The 95% CI of 0.02–0.25 around the primary HR shows the range of hazard-ratio values supported by the analysis under its statistical assumptions and sampling framework. Confidence intervals are therefore essential for understanding how precisely the treatment effect was estimated.
Why can a zero event percentage still have a nonzero confidence interval?
The dabigatran unexplained-death estimate is 0.0, with a 95% Clopper-Pearson interval of 0.00–0.54. A zero observed count does not imply that the underlying event probability is known to be exactly zero. The exact binomial interval reflects the uncertainty that remains when no events are observed.
Why use Fisher exact testing?
Fisher exact testing is designed for categorical two-group comparisons and remains applicable when event counts are small. That makes it a natural method for several of the rare-event secondary endpoints in this registry analysis, including unexplained death and PE.
Why do the efficacy and bleeding analyses use different analysis populations?
The primary VTE analysis was performed in the full analysis set and analyzed as randomized, preserving the randomized treatment comparison. The bleeding analyses were performed in the full analysis set and analyzed as treated, reflecting actual treatment received. The registry specifically notes 3 participants randomized to placebo but treated with dabigatran, making the distinction operationally relevant.
Why should the hazard ratio and event percentage not be treated as interchangeable?
An event percentage answers a binary question over a specified period: what proportion experienced the endpoint? A hazard ratio is a relative time-to-event measure that incorporates event timing and censoring. Two analyses can therefore describe the same clinical domain while answering different statistical questions.
14. Confidence Intervals and Small Event Counts
Several secondary endpoints have low observed percentages, making the choice of confidence-interval method especially relevant. The registry reports exact Clopper-Pearson two-sided 95% confidence intervals for these group-specific percentages.
| Endpoint | Dabigatran | 95% CI | Placebo | 95% CI |
|---|---|---|---|---|
| DVT | 0.3 | 0.04–1.06 | 3.5 | 2.21–5.17 |
| PE | 0.1 | 0.00–0.82 | 2.1 | 1.16–3.52 |
| Unexplained death | 0.0 | 0.00–0.54 | 0.3 | 0.04–1.09 |
| Bleeding event | 0.3 | 0.04–1.05 | 0.0 | 0.00–0.56 |
These intervals illustrate an important principle in clinical biostatistics: a small observed event proportion does not automatically imply a highly precise estimate. When events are uncommon, uncertainty can remain substantial even in a relatively large treatment group.
15. Reading the P-values Correctly
P < 0.0001
The reported primary superiority analysis provides very strong statistical evidence against its null hypothesis under the specified Cox-model framework.
P = 0.2428
The unexplained-death Fisher exact comparison does not provide conventional statistical evidence against its null hypothesis.
P = 0.0013
The reported time-to-first MBE or CRBE analysis provides statistical evidence against its null hypothesis under the Cox framework.
P = 0.4998
The binary bleeding Fisher exact comparison does not provide conventional statistical evidence against its null hypothesis.
These P-values cannot be placed on a single scale of “effectiveness.” A P-value is determined jointly by the observed data, the null hypothesis, and the statistical model or test. Effect magnitude and precision require the corresponding effect estimate and confidence interval.
16. Interpreting the Primary Hazard Ratio in Context
The primary HR of 0.08 represents a large relative separation in estimated time-to-event hazard between dabigatran and placebo for the centrally confirmed symptomatic recurrent VTE composite including unexplained death.
The 95% CI of 0.02–0.25 is relatively far from the null value of 1 throughout its range. It still represents uncertainty: the point estimate is 0.08, but the interval indicates that the underlying hazard ratio is not known with perfect precision.
The hazard ratio alone does not provide the absolute number or percentage of recurrent VTE events prevented. Absolute event probabilities require corresponding event counts and denominators or survival estimates at specified time points.
The statistical result establishes the direction and magnitude of the reported time-to-event association in the randomized comparison. Clinical interpretation also requires considering the separate bleeding and safety findings rather than evaluating the efficacy HR in isolation.
17. Efficacy and Bleeding Should Be Interpreted Separately
RE-SONATE illustrates why a clinical trial cannot be reduced to one P-value or one hazard ratio. The primary VTE endpoint and bleeding endpoints address different outcomes and point in different statistical directions.
| Domain | Reported analysis | Estimate | Interpretation of direction |
|---|---|---|---|
| Primary recurrent VTE including unexplained death | Cox model | HR 0.08 | Lower estimated VTE hazard with dabigatran |
| Recurrent VTE excluding unexplained death | Cox model | HR 0.08 | Lower estimated VTE hazard with dabigatran |
| MBE or CRBE | Cox model | HR 2.92 | Higher estimated bleeding-event hazard with dabigatran |
| Any bleeding event | Cox model | HR 1.82 | Higher estimated bleeding-event hazard with dabigatran |
The statistical interpretation is therefore multidimensional. The efficacy endpoint has a hazard ratio below 1, while the reported bleeding time-to-event analyses have hazard ratios above 1. These are separate estimands for separate clinical outcomes.
18. Why This Trial Matters Statistically
RE-SONATE is a useful teaching example because the registry results bring together randomized treatment comparison, double blinding, a composite time-to-event primary endpoint, Cox regression, exact binomial confidence intervals, Fisher exact testing, and a deliberate distinction between as-randomized efficacy analysis and as-treated bleeding analysis.
| Concept | How it appears in RE-SONATE |
|---|---|
| Randomization | Participants were randomized to dabigatran or matching placebo. |
| Blinding | The trial was double blind. |
| Time-to-event endpoint | The primary endpoint was centrally confirmed symptomatic recurrent VTE including unexplained death during 6 months. |
| Cox proportional-hazards model | Used for the primary endpoint and reported bleeding time-to-event analyses. |
| Hazard ratio | Used as the effect measure for the Cox analyses. |
| Confidence intervals | 95% intervals were reported for the primary and secondary Cox hazard ratios and for binary event percentages. |
| Exact binomial analysis | Clopper-Pearson intervals were used for several rare-event percentages. |
| Fisher exact test | Used for DVT, PE, unexplained death, and bleeding binary comparisons. |
| Intention-to-treat principle | The primary efficacy analysis was performed in the FAS and analyzed as randomized. |
| As-treated safety analysis | Bleeding analyses used FAS and analyzed as treated. |
19. Important Limitations and Interpretation Issues
- Hazard-ratio interpretation: a Cox HR is a model-based relative measure and should not be interpreted as an absolute risk reduction or as the percentage of participants benefiting.
- Proportional-hazards assumption: the Cox model relies on a proportional-hazards framework. The registry information reported here does not provide a separate assessment of that assumption.
- Rare events: several secondary outcomes have very low observed percentages, so group-specific confidence intervals remain important even when the observed percentage is small.
- Different analysis populations: efficacy was analyzed as randomized, while bleeding analyses were analyzed as treated. These analyses therefore have different interpretations.
- Three treatment-crossed participants: the registry identifies 3 participants randomized to placebo but treated with dabigatran in the bleeding analysis population.
- Composite endpoint: the primary endpoint combines recurrent DVT, fatal or non-fatal PE, and unexplained death according to the registry definition. A composite result does not imply that every component has the same effect size.
- Multiple secondary analyses: the registry reports multiple secondary statistical analyses. Individual P-values should therefore be interpreted in the context of the overall analysis plan rather than treated as isolated evidence without considering multiplicity.
- Absolute effects: the posted primary Cox analysis provides a hazard ratio and confidence interval, but the hazard ratio alone does not supply an absolute risk difference.
20. A Practical Framework for Reading RE-SONATE
Step 1: Identify the estimand
The primary estimand is a time-to-event comparison of centrally confirmed symptomatic recurrent VTE including unexplained death during the intended treatment period.
Step 2: Identify the effect measure
The primary effect measure is a hazard ratio, not a risk ratio or percentage-point difference.
Step 3: Read the interval
The 95% CI of 0.02–0.25 shows the statistical precision of the primary hazard-ratio estimate.
Step 4: Read the P-value
P < 0.0001 provides evidence against the primary null hypothesis but does not measure effect magnitude.
Step 5: Separate outcomes
VTE efficacy and bleeding are different endpoints and should not be combined into one statistical conclusion.
Step 6: Check the population
The primary efficacy analysis is as randomized, while bleeding analyses are as treated.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Calculators
23. Sources
- ClinicalTrials.gov: NCT00558259 — RE-SONATE.
- Publication: PubMed record for PMID 24081972.
- Publication: PubMed record for PMID 23425163.
Continue through Clinical Biostats
Connect the endpoints and statistical methods used in RE-SONATE with deeper biostatistics tutorials and practical statistical calculators.
24. Record Summary
RE-SONATE provides a clear example of how a randomized clinical trial can combine time-to-event and categorical statistical methods to evaluate both efficacy and safety. The primary analysis used a Cox proportional-hazards model in the full analysis set analyzed as randomized and reported an HR of 0.08 (95% CI 0.02–0.25; P < 0.0001) for centrally confirmed symptomatic recurrent VTE including unexplained death during the intended 6-month treatment period.
The secondary analyses extend the statistical picture. Recurrent VTE excluding unexplained death produced an HR of 0.08 (95% CI 0.03–0.27; P < 0.0001). Binary DVT and PE analyses used exact confidence intervals and Fisher exact testing, while unexplained death had a Fisher exact P-value of 0.2428. Bleeding analyses used the as-treated population and produced Cox hazard ratios of 2.92 for time to first MBE or CRBE and 1.82 for time to first bleeding event.
The central statistical lesson is that these numbers answer different questions. The primary HR describes a relative time-to-event effect; the Clopper-Pearson intervals describe uncertainty around individual event percentages; Fisher exact P-values test categorical group differences; and the bleeding Cox models describe separate time-to-event safety outcomes. Reading the trial correctly therefore requires keeping the endpoint, estimand, analysis population, effect measure, confidence interval, and hypothesis test aligned.