This page separates reported trial results from statistical interpretation. Numerical results and trial characteristics are restricted to the ClinicalTrials.gov record for NCT02239120. The ClinicalTrials.gov record provides the official trial registry record.
1. Trial at a Glance
RE-SPECT ESUS was a randomized, double-masked, phase 3 trial in neurology and secondary stroke prevention. The trial enrolled 5390 participants and compared dabigatran etexilate 110 or 150 mg with acetylsalicylic acid (aspirin) 100 mg using time-to-event analyses based on Cox proportional-hazards models.
| Feature | RE-SPECT ESUS |
|---|---|
| Trial name | RE-SPECT ESUS |
| Phase | Phase 3 |
| Therapeutic area | Neurology |
| Conditions | Stroke; Secondary Prevention |
| Allocation | Randomized |
| Design model | Single group, as recorded in the registry design field |
| Masking | Double |
| Primary purpose | Prevention |
| Enrollment | 5390 |
| Arms | 2 |
| Lead sponsor | Boehringer Ingelheim |
| Sponsor type | Industry |
| Status | Completed |
| Start | 2014-11-27 |
| Primary completion | 2018-08-14 | Results posted | Yes |
| Outcome measures posted | 10 |
| Statistical analyses posted | 9 |
The registry data contain an apparent tension between the randomized allocation and two-arm structure and the recorded design-model field of SINGLE_GROUP. This page reports both fields rather than silently resolving the discrepancy. The substantive treatment comparison in the posted analyses is explicitly between dabigatran etexilate 110 or 150 mg and acetylsalicylic acid 100 mg.
2. Clinical Question
The trial addresses whether dabigatran etexilate, compared with acetylsalicylic acid (aspirin) 100 mg, differs in the rate of adjudicated recurrent stroke among patients being studied for secondary prevention after embolic stroke of undetermined source.
Population
Patients represented by the trial's brief title as having embolic stroke of undetermined source and being treated for secondary stroke prevention.
Intervention
Dabigatran etexilate 110 or 150 mg. The intervention list also records optional ASA as a comedication and placebo components.
Comparator
Acetylsalicylic acid (aspirin) 100 mg, with the intervention list also recording placebo to dabigatran and placebo to optional ASA as comedication.
Primary question
How does dabigatran etexilate compare with aspirin for the registered primary efficacy endpoint of adjudicated recurrent stroke, and what is the corresponding comparison for first major bleed as the primary safety endpoint?
3. Trial Design
The registry profile describes a randomized, double-masked, phase 3 prevention trial with 5390 enrolled participants and two treatment arms. The posted primary analyses use survival-analysis methods and hazard ratios rather than simple proportions at a fixed time point.
Dabigatran etexilate
- Dabigatran etexilate 110 or 150 mg
- Optional ASA as comedication is included in the intervention listing
- Placebo to optional ASA as comedication is also recorded
Acetylsalicylic acid
- Acetylsalicylic acid (aspirin) 100 mg
- Placebo to dabigatran etexilate is included in the intervention listing
- Placebo to optional ASA as comedication is also recorded
The ClinicalTrials.gov record does not provide a randomization ratio, arm-specific randomized sample sizes, crossover information, or a detailed treatment schedule. Those design features are therefore not reconstructed here.
4. Endpoints
The registry identifies two primary endpoints, both represented as time-to-event outcomes in the posted statistical analyses. The primary endpoint types are recorded as count/rate, with annualised event rate expressed as % per year in the statistical-analysis records.
| Endpoint | Registered time frame | Role | Analysis population |
|---|---|---|---|
| Adjudicated Recurrent Stroke | From randomisation until full follow up period, approximately 43 months. | Primary efficacy | Randomised set (RS) |
| First Major Bleed (Adjudicated) | Between the first trial medication intake up to 6 days after the last trial medication intake, approximately 42 months. | Primary safety | Treated set (TS) |
Adjudicated Recurrent Stroke
The registry definition states that adjudicated recurrent stroke includes ischemic, hemorrhagic, or unspecified stroke. The annualised event rate represents the average number of events per patient during a 1-year period.
First Major Bleed (Adjudicated)
5. Analysis Populations
The two primary endpoints use different analysis populations. That distinction is central to interpreting the hazard ratios because the populations define who contributes to each analysis and when observation begins.
| Population | Registry definition / role |
|---|---|
| Randomised set (RS) | Consisted of all participants who were randomised, regardless of whether they took trial medication. Used for the adjudicated recurrent-stroke primary analysis and the other efficacy analyses recorded as RS. |
| Treated set (TS) | Consisted of all patients who were treated with at least 1 dose of trial medication. Used for the first-major-bleed primary safety analysis and the other safety analyses recorded as TS. |
This difference illustrates a common principle in clinical-trial statistics: efficacy and safety questions can use different analysis populations because their estimands and observation windows are different. It also means the two primary hazard ratios should not be treated as though they were calculated from exactly the same patient set.
6. Statistical Methodology
Cox proportional-hazards model
All nine registry-reported statistical analyses use a regression method recorded as Cox, normalized to a Cox proportional-hazards model. The method category is survival analysis, and the effect measure is the hazard ratio.
The model relates the hazard at time t to baseline hazard and covariates. For a treatment indicator, the exponentiated treatment coefficient is interpreted as a hazard ratio under the proportional-hazards framework.
Covariate adjustment
The posted analyses specify adjustment for three covariate categories: age (≥75 years versus <75 years), creatinine clearance (≥50 mL/min versus <50 mL/min), and stroke or transient ischaemic attack (TIA) prior to the index stroke.
| Covariate | Registry categorization |
|---|---|
| Age | ≥75 years versus <75 years |
| Creatinine clearance | ≥50 mL/min versus <50 mL/min |
| Prior cerebrovascular event | Stroke or TIA prior to index stroke |
Hazard ratio
The hazard ratio is a relative time-to-event measure. An HR below 1 indicates a lower estimated hazard in the dabigatran group relative to aspirin under the fitted model; an HR above 1 indicates a higher estimated hazard.
HR < 1 → lower modeled hazard with dabigatran
HR > 1 → higher modeled hazard with dabigatran
The hazard ratio is not an absolute event probability and does not directly state how many individual patients benefited or experienced harm.
7. Statistical Methods Explained
Why use a Cox proportional-hazards model?
The primary outcomes are time-to-event endpoints rather than simple binary outcomes at a single fixed time. Cox regression uses information about when events occur and can accommodate right-censored follow-up. It also permits adjustment for the three covariate categories specified in the registry analysis notes.
What does an HR of 0.85 mean for recurrent stroke?
An HR of 0.85 means the fitted model estimates the hazard of adjudicated recurrent stroke in the dabigatran group at 85% of the corresponding hazard in the aspirin group. Equivalently, 1 − 0.85 = 0.15, so the model-based relative hazard is approximately 15% lower. This is a statement about the estimated hazard, not a statement that 15% of patients avoided stroke.
What does an HR of 1.36 mean for major bleeding?
An HR of 1.36 means the fitted model estimates the hazard of first major bleed in the dabigatran group at 136% of the corresponding hazard in the aspirin group. The corresponding relative increase in modeled hazard is 36%. Again, this is not a 36-percentage-point increase in the probability of bleeding.
Why does the analysis adjust for covariates?
The registry explicitly identifies age, creatinine clearance, and prior stroke or TIA before the index stroke as covariates. Adjustment incorporates these prespecified variables into the regression model, allowing the reported treatment hazard ratio to account for their modeled contribution rather than relying solely on an unadjusted comparison.
Why is the confidence interval important?
A point estimate such as HR 0.85 is only one estimate of the treatment effect. Its 95% confidence interval of 0.69 to 1.03 communicates statistical uncertainty around that estimate under the model and sampling framework. A wider interval indicates less precision than a narrower interval; the interval should not be interpreted as a range containing the effects experienced by individual patients.
Why does the p-value not measure effect size?
The p-value addresses compatibility with the tested null hypothesis under the specified statistical framework. It does not measure the magnitude or clinical importance of an effect. Effect size is conveyed by the hazard ratio, while the confidence interval conveys uncertainty around that effect estimate.
Why does the proportional-hazards assumption matter?
The Cox hazard ratio is most naturally interpreted as a relative hazard under the proportional-hazards framework. If the relative hazard changes substantially over time, a single HR can compress a more complicated time-varying pattern into one number. The ClinicalTrials.gov record does not report a formal proportional-hazards diagnostic, so the reported HR should be understood within the assumptions of the model used.
8. Primary Result: Adjudicated Recurrent Stroke
The primary recurrent-stroke analysis compared dabigatran etexilate 110 or 150 mg with acetylsalicylic acid 100 mg from randomisation through the full follow-up period, approximately 43 months. The analysis population was the randomised set.
Hazard ratio for adjudicated recurrent stroke
95% CI: 0.69–1.03 · P = 0.1028
Cox proportional-hazards model with covariate adjustment.
| Primary endpoint | Dabigatran vs ASA | 95% CI | P-value | Population |
|---|---|---|---|---|
| Adjudicated Recurrent Stroke | HR 0.85 | 0.69–1.03 | 0.1028 | Randomised set |
What the estimate means: the adjusted hazard ratio of 0.85 corresponds to an estimated hazard of recurrent stroke that is approximately 15% lower with dabigatran than with aspirin under the fitted Cox model.
What it does not mean: HR 0.85 does not mean that 15% of participants avoided recurrent stroke, nor does it represent a 15-percentage-point difference in cumulative stroke probability.
What the confidence interval says: the 95% CI of 0.69–1.03 includes 1.00. Thus, the registry-reported estimate is compatible with a range extending from a lower modeled hazard to a value slightly above equal hazard. The interval is therefore important for understanding the uncertainty surrounding the point estimate.
Why the p-value is not an effect-size measure: P = 0.1028 is a test result under the superiority framework recorded by the registry. It does not say that the effect is "10.28%" in size and should not replace the HR and confidence interval when describing the magnitude and precision of the comparison.
Analysis caution: this result is based on the randomised set and the registered follow-up window. It is also model-based, so its interpretation depends on the Cox-model framework and the proportional-hazards assumption.
9. Primary Safety Result: First Major Bleed
The primary safety endpoint was first major bleed, analyzed between first trial medication intake and 6 days after the last trial medication intake, approximately 42 months. The analysis population was the treated set.
Hazard ratio for first major bleed
95% CI: 0.94–1.97 · P = 0.1076
Cox proportional-hazards model with covariate adjustment.
| Primary safety endpoint | Dabigatran vs ASA | 95% CI | P-value | Population |
|---|---|---|---|---|
| First Major Bleed (Adjudicated) | HR 1.36 | 0.94–1.97 | 0.1076 | Treated set |
What the estimate means: the adjusted HR of 1.36 estimates a 36% higher hazard of first major bleed with dabigatran relative to aspirin under the fitted Cox model.
What it does not mean: it does not mean that 36% of dabigatran-treated participants experienced a major bleed, or that the absolute probability of bleeding was 36 percentage points higher.
What the confidence interval says: the 95% CI of 0.94–1.97 spans 1.00. The point estimate therefore carries substantial uncertainty, with the interval compatible with a modeled hazard somewhat below equal hazard as well as a substantially higher hazard.
Why the p-value is not the effect size: P = 0.1076 is the reported superiority-test result. It does not quantify the clinical magnitude of bleeding risk. The HR and confidence interval provide the effect estimate and its precision.
Analysis caution: this safety analysis uses the treated set rather than the randomised set, and its observation window begins with first medication intake rather than randomisation. These design details should be retained when comparing the safety result with the recurrent-stroke efficacy result.
10. Secondary Endpoint Results
The registry contains seven secondary statistical analyses in the ClinicalTrials.gov record: five efficacy or mortality outcomes analyzed in the randomised set and three bleeding outcomes analyzed in the treated set. All use Cox proportional-hazards regression with the same three categories of covariate adjustment.
| Secondary endpoint | Analysis population | HR | 95% CI | P-value |
|---|---|---|---|---|
| Adjudicated Ischaemic Stroke | RS | 0.84 | 0.68–1.03 | 0.0892 |
| Adjudicated Composite of Non-fatal Stroke, Non-fatal Myocardial Infarction, or Cardiovascular Death | RS | 0.88 | 0.73–1.06 | 0.1911 |
| Disabling Stroke | RS | 0.59 | 0.36–0.96 | 0.0354 |
| All-cause Death | RS | 0.96 | 0.66–1.38 | 0.8074 |
| Adjudicated Intracranial Hemorrhage | TS | 1.03 | 0.58–1.83 | 0.9064 |
| Adjudicated Life-threatening Bleed | TS | 0.82 | 0.49–1.36 | 0.4352 |
| Any Bleed (Investigator-reported) | TS | 1.28 | 1.12–1.47 | 0.0003 |
Adjudicated Ischaemic Stroke
The estimate corresponds to an approximately 16% lower modeled hazard with dabigatran relative to aspirin, but the confidence interval extends through 1.00. This was analyzed in the randomised set from randomisation until the full follow-up period, up to 43 months.
Composite of Non-fatal Stroke, Non-fatal Myocardial Infarction, or Cardiovascular Death
The model estimates a 12% lower hazard for the composite with dabigatran. The confidence interval includes 1.00, so the point estimate should not be interpreted independently of its uncertainty.
Disabling Stroke
The estimated hazard of disabling stroke was 41% lower with dabigatran under the fitted model. The 95% confidence interval remains below 1.00, although its width indicates appreciable uncertainty in the magnitude of the estimate.
All-cause Death
The point estimate is close to 1.00, corresponding to an estimated 4% lower hazard with dabigatran. The wide confidence interval indicates substantial uncertainty and includes both lower and higher modeled hazards.
Adjudicated Intracranial Hemorrhage
The point estimate is close to 1.00, but the confidence interval is wide. This means the estimate alone should not be interpreted as establishing equivalence or absence of an important difference.
Adjudicated Life-threatening Bleed
The point estimate corresponds to an approximately 18% lower modeled hazard with dabigatran, but the confidence interval includes 1.00 and extends over a relatively broad range.
Any Bleed (Investigator-reported)
Hazard ratio for any bleed
95% CI: 1.12–1.47 · P = 0.0003
Treated-set analysis from first trial medication intake to 6 days after the last trial medication intake, approximately 42 months.
The any-bleed analysis estimates a 28% higher modeled hazard with dabigatran relative to aspirin. Unlike several other secondary endpoints, its 95% CI of 1.12–1.47 remains above 1.00, and the reported P-value is 0.0003.
That does not mean that 28% more patients bled, nor does it establish the absolute difference in bleeding probability. It also does not automatically determine the interpretation of the primary major-bleed endpoint: any bleed and first major bleed are different endpoints, and the former is investigator-reported whereas the latter is adjudicated.
The analysis is also based on the treated set and the medication-exposure window, so its analysis population and time origin differ from the primary recurrent-stroke analysis.
11. Safety Results
The ClinicalTrials.gov record provides serious adverse-event counts by treatment arm. These figures are reported as affected participants divided by participants at risk.
| Safety measure | Dabigatran Etexilate 110 or 150 mg | Acetylsalicylic Acid, Aspirin 100 |
|---|---|---|
| Serious adverse events | 724 / 2676 | 740 / 2674 |
The counts above are not themselves a time-to-event comparison and should not be substituted for the adjudicated major-bleed Cox analysis. The registry separately reports a treated-set hazard ratio for first major bleed and a treated-set hazard ratio for any investigator-reported bleed.
12. Primary and Secondary Results in Context
Viewed together, the posted analyses produce a heterogeneous set of hazard-ratio estimates rather than a single summary measure. The primary recurrent-stroke HR is 0.85, the primary major-bleed HR is 1.36, and secondary estimates range from 0.59 for disabling stroke to 1.28 for any investigator-reported bleed.
| Outcome family | Direction of point estimate | Key statistical feature |
|---|---|---|
| Adjudicated recurrent stroke | HR 0.85 | 95% CI 0.69–1.03 |
| First major bleed | HR 1.36 | 95% CI 0.94–1.97 |
| Ischaemic stroke | HR 0.84 | 95% CI 0.68–1.03 |
| Composite vascular endpoint | HR 0.88 | 95% CI 0.73–1.06 |
| Disabling stroke | HR 0.59 | 95% CI 0.36–0.96 |
| All-cause death | HR 0.96 | 95% CI 0.66–1.38 |
| Intracranial hemorrhage | HR 1.03 | 95% CI 0.58–1.83 |
| Life-threatening bleed | HR 0.82 | 95% CI 0.49–1.36 |
| Any investigator-reported bleed | HR 1.28 | 95% CI 1.12–1.47 |
This pattern demonstrates why clinical-trial interpretation should not be reduced to asking whether every individual p-value is below a threshold. Endpoints differ in definition, population, time origin, event frequency, adjudication process, and clinical meaning. The primary efficacy and primary safety endpoints also address different questions.
13. Confidence Intervals and the Meaning of Precision
The confidence intervals vary substantially in width. This is particularly visible for outcomes such as intracranial hemorrhage and all-cause death, where the intervals extend across a broad range around the point estimate.
Recurrent stroke
HR 0.85 with 95% CI 0.69–1.03. The interval crosses 1.00, so uncertainty around the estimate includes both lower and slightly higher modeled hazard.
Major bleed
HR 1.36 with 95% CI 0.94–1.97. The point estimate is above 1.00, but the interval includes 1.00 and extends to 1.97.
Disabling stroke
HR 0.59 with 95% CI 0.36–0.96. The interval is entirely below 1.00, although the lower and upper bounds still represent materially different effect magnitudes.
Intracranial hemorrhage
HR 1.03 with 95% CI 0.58–1.83. The point estimate is near 1.00, but the interval is wide enough that substantial differences remain compatible with the estimate.
A confidence interval should therefore be read as part of the estimate rather than as an accessory to a p-value. In a time-to-event trial, the combination of HR + confidence interval + endpoint definition + analysis population + time window is much more informative than any one number alone.
14. Time Frames and Censoring
Time-to-event analysis differs from a fixed-time proportion because not every participant necessarily contributes the same amount of observable follow-up. The Cox framework can incorporate participants whose event status is not observed for the entire nominal follow-up period by using their information up to the relevant censoring time.
| Endpoint group | Time origin | Follow-up window | Population |
|---|---|---|---|
| Recurrent stroke and other RS outcomes | Randomisation | Full follow-up period, approximately 43 months / up to 43 months | Randomised set |
| Major bleed and other TS bleeding outcomes | First trial medication intake | Approximately 42 months; up to 6 days after last trial medication intake | Treated set |
This distinction is statistically important. A hazard ratio is always tied to a defined time origin, event definition, censoring framework, and analysis population. Changing any of these can change the estimand being studied.
15. Superiority Testing
The statistical analyses posted on ClinicalTrials.gov are all identified as having a superiority hypothesis type. This means the statistical question is framed around whether the treatment groups differ, rather than whether one treatment can be shown to be no worse than another within a prespecified non-inferiority margin.
This is a conceptual representation of a two-sided hazard-ratio comparison. The registry specifically identifies the posted analyses as superiority analyses and reports two-sided 95% confidence intervals.
The ClinicalTrials.gov record does not provide a non-inferiority margin, crossover specification, factorial design, Bayesian analysis, interim-analysis rule, or missing-data/imputation method. Those topics are therefore not incorporated into the trial's statistical description here.
16. Why the Analysis Population Matters
One of the most instructive aspects of RE-SPECT ESUS is that the primary efficacy and primary safety analyses do not use the same analysis population.
The difference preserves an important distinction between randomized treatment comparison and treatment-exposure safety assessment. It also means that directly comparing the numerical HRs from the two primary endpoints as though they were estimates from a common population would be misleading.
17. Interpreting the Primary Endpoint Pair
The two primary endpoints can be viewed as addressing complementary dimensions of the treatment comparison: adjudicated recurrent stroke is the primary efficacy endpoint, while first major bleed is the primary safety endpoint.
Efficacy question
Does dabigatran change the hazard of adjudicated recurrent stroke relative to aspirin during the registered efficacy follow-up?
Safety question
Does dabigatran change the hazard of a first adjudicated major bleed during the registered medication-exposure safety window?
Different time origins
The stroke endpoint begins at randomisation, whereas the major-bleed endpoint begins at first trial medication intake.
Different populations
The recurrent-stroke analysis uses the randomised set; the major-bleed analysis uses the treated set.
This structure is statistically preferable to collapsing efficacy and safety into a single informal "net benefit" number that was not reported in the registry analyses.
18. Limitations
- Registry design-field inconsistency: the registry-reported profile records allocation as randomized and identifies two arms, while the design-model field is recorded as SINGLE_GROUP. This page reports the fields as reported in the registry rather than resolving the discrepancy without source evidence.
- No arm-specific randomized sample sizes: the ClinicalTrials.gov record provides total enrollment of 5390 but do not provide the number randomized to each arm.
- Different analysis populations: the primary efficacy endpoint uses the randomised set, while the primary safety endpoint uses the treated set.
- Different time origins: efficacy follow-up begins at randomisation, whereas safety follow-up begins at first medication intake.
- Cox-model assumptions: hazard ratios are model-based and depend on the proportional-hazards framework. The ClinicalTrials.gov record does not report a formal diagnostic of that assumption.
- Multiplicity: the ClinicalTrials.gov record identifies two primary endpoints and multiple secondary analyses, but do not provide the detailed multiplicity-control strategy. Individual secondary p-values should therefore be interpreted in the context of their endpoint hierarchy rather than as isolated findings.
- Secondary endpoints: secondary analyses can provide supportive information, but they do not automatically carry the same confirmatory role as primary endpoints.
- Safety endpoint definitions: major bleed, intracranial hemorrhage, life-threatening bleed, and any bleed are distinct outcomes and should not be conflated.
- Summary-level reporting: the ClinicalTrials.gov record contains hazard ratios, confidence intervals, and p-values but do not provide underlying individual-level event and censoring times needed to reconstruct Kaplan-Meier curves.
- Unreported design details: the ClinicalTrials.gov record does not provide information sufficient to characterize crossover, interim monitoring, Bayesian methods, or a specific missing-data/imputation strategy.
19. Why This Trial Matters Statistically
RE-SPECT ESUS is a useful teaching case because it demonstrates how a randomized clinical trial can generate several distinct statistical questions from the same treatment comparison. The core analysis is not simply "did more or fewer patients have a stroke?" Instead, the registry uses time-to-event methods, model-based adjustment, different analysis populations, and different observation windows for efficacy and safety.
| Concept | How it appears in RE-SPECT ESUS |
|---|---|
| Randomization | The trial allocation is recorded as randomized, with two treatment arms. |
| Double masking | The registry records double masking and placebo components among the interventions. |
| Time-to-event endpoints | Primary and posted secondary analyses are analyzed using Cox regression. |
| Cox proportional hazards | The normalized statistical method for all registry-reported formal analyses. |
| Hazard ratio | The reported effect measure for the primary and secondary time-to-event analyses. |
| Covariate adjustment | Models adjust for age, creatinine clearance, and prior stroke/TIA before the index stroke. |
| Confidence intervals | All registry-reported formal analyses report two-sided 95% confidence intervals. |
| Superiority testing | All registry-reported statistical analyses are classified as superiority analyses. |
| Different populations | Recurrent stroke uses the randomised set; bleeding analyses use the treated set. |
| Different time origins | Efficacy begins at randomisation; safety begins at first medication intake. |
| Endpoint hierarchy | Two primary endpoints are followed by multiple secondary outcomes. |
20. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary recurrent-stroke analysis produced HR 0.85 with 95% CI 0.69–1.03 and P = 0.1028. The primary major-bleed analysis produced HR 1.36 with 95% CI 0.94–1.97 and P = 0.1076. Both were Cox analyses with covariate adjustment and superiority hypotheses.
Clinical interpretation
The numerical treatment effects must be considered endpoint by endpoint. Recurrent stroke, disabling stroke, major bleeding, intracranial hemorrhage, and any bleeding represent different clinical outcomes and have different statistical estimates in the registry.
A useful clinical-statistical reading therefore avoids treating a single HR as a complete description of the trial. The appropriate unit of interpretation is the specific endpoint and its prespecified analysis.
21. A Closer Look at the Hazard Ratio
HR 0.85 means the estimated instantaneous hazard under the Cox model is 15% lower in the dabigatran group than in the aspirin group. The 95% CI of 0.69–1.03 shows that the precision around this estimate includes values below and slightly above 1.00.
HR 1.36 means the estimated instantaneous hazard under the Cox model is 36% higher in the dabigatran group than in the aspirin group. The 95% CI of 0.94–1.97 shows that the uncertainty includes 1.00 and extends to a considerably higher hazard.
HR 1.28 means the estimated instantaneous hazard of any investigator-reported bleed is 28% higher with dabigatran under the fitted model. Its 95% CI of 1.12–1.47 lies above 1.00, but this endpoint is not interchangeable with the primary adjudicated major-bleed endpoint.
These examples show why the phrase "lower risk" should be used carefully. A hazard ratio describes a modeled relative rate over time; it does not by itself supply the absolute number of events prevented or caused.
22. What the P-values Tell Us
The analyses posted on ClinicalTrials.gov report p-values ranging from 0.0003 for any investigator-reported bleed to 0.9064 for adjudicated intracranial hemorrhage. These numbers should be read alongside their corresponding effect estimates and confidence intervals.
| Endpoint | P-value | Why the HR and CI remain necessary |
|---|---|---|
| Adjudicated recurrent stroke | 0.1028 | HR 0.85 and CI 0.69–1.03 describe magnitude and uncertainty. |
| First major bleed | 0.1076 | HR 1.36 and CI 0.94–1.97 describe the estimated direction and precision. |
| Disabling stroke | 0.0354 | HR 0.59 and CI 0.36–0.96 quantify the estimated relative effect. |
| Any bleed | 0.0003 | HR 1.28 and CI 1.12–1.47 show the estimated magnitude and precision. |
| Intracranial hemorrhage | 0.9064 | HR 1.03 and CI 0.58–1.83 show that the point estimate is imprecise. |
A p-value is not a probability that the null hypothesis is true, nor is it a measure of the clinical importance of the result. The confidence interval is essential for understanding what effect sizes remain compatible with the observed data under the model.
23. Related Statistical Concepts
Learn more about the methods used in this trial:
24. Related Statistical Calculators
25. Sources
- ClinicalTrials.gov: NCT02239120 — RE-SPECT ESUS.
- Linked PubMed record: PMID 35656979.
- Linked PubMed record: PMID 35538232.
- Linked PubMed record: PMID 34649459.
- Linked PubMed record: PMID 33588594.
- Linked PubMed record: PMID 33504190.
Continue through the Clinical Biostats statistical pathway
Explore the statistical concepts behind randomized trials, time-to-event endpoints, Cox regression, hazard ratios, confidence intervals, and related analysis methods.
26. Record Summary
RE-SPECT ESUS provides a useful statistical case study in randomized time-to-event analysis. The registry records a phase 3, double-masked, randomized trial with 5390 enrolled participants and two arms. Its two primary endpoints use Cox proportional-hazards models with hazard ratios and two-sided 95% confidence intervals, while the primary efficacy and safety analyses use different populations and different time origins. The primary recurrent-stroke estimate was HR 0.85 (95% CI 0.69–1.03; P = 0.1028), while the primary first-major-bleed estimate was HR 1.36 (95% CI 0.94–1.97; P = 0.1076).
The secondary analyses further illustrate why endpoint-specific interpretation matters. Estimates ranged from HR 0.59 for disabling stroke to HR 1.28 for any investigator-reported bleed, with materially different confidence intervals, p-values, endpoint definitions, and analysis populations. The most informative reading therefore combines the hazard ratio, the confidence interval, the p-value, the endpoint definition, the time origin, and the analysis population rather than relying on any single statistic.