This page provides an independent statistical analysis and educational interpretation of publicly reported results. It describes the registered trial design, endpoint definition, and statistical framework for the INSPIRES study. ClinicalTrials.gov provides the official trial registry record. View the ClinicalTrials.gov record.
1. Trial at a Glance
INSPIRES is a randomized, quadruple-masked, phase 3 factorial trial evaluating intensive medical therapy for high-risk intracranial or extracranial atherosclerosis in patients with acute stroke or transient ischemic attack. The registry reports 6100 participants and four study arms, with stroke at 90 days designated as the primary endpoint.
| Feature | INSPIRES |
|---|---|
| Trial acronym | INSPIRES |
| NCT identifier | NCT03635749 |
| Title | Intensive Medical Therapy for High-risk Intracranial or Extracranial Atherosclerosis |
| Phase | Phase 3 |
| Status | Unknown |
| Conditions | Acute Stroke; Transient Ischemic Attack |
| Allocation | Randomized |
| Design model | Factorial |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 6100 |
| Number of arms | 4 |
| Lead sponsor | Beijing Tiantan Hospital |
| Sponsor type | Other |
2. Clinical Question
The registered study asks a treatment question in patients with acute stroke or transient ischemic attack in the setting of high-risk intracranial or extracranial atherosclerosis. Its factorial structure evaluates two medical-therapy dimensions: the intensity of antiplatelet treatment and the timing of high-intensity statin treatment.
Population
Patients with acute stroke or transient ischemic attack and high-risk intracranial or extracranial atherosclerosis.
Antiplatelet intervention
Intensive antiplatelet drug therapy.
Antiplatelet comparator
Standard antiplatelet drug therapy.
Statin intervention
Immediate high-intensity statin drug therapy.
The registry also identifies delayed high-intensity statin treatment as the corresponding statin comparator. Because the design is factorial with four arms, the treatment structure permits evaluation of these two therapy dimensions within a single randomized trial.
3. Trial Design
Factorial structure
A factorial design studies more than one treatment dimension simultaneously. Here, one dimension concerns antiplatelet intensity and the other concerns the timing of high-intensity statin therapy. With two levels for each dimension, the design produces four treatment combinations.
Antiplatelet strategy
- Intensive antiplatelet drug therapy
- Standard antiplatelet drug therapy
Statin timing
- Immediate high-intensity statin drug therapy
- Delayed high-intensity statin drug therapy
The factorial structure is statistically important because it can estimate treatment effects for each factor while using the information generated across the full trial population. If the effects of the two factors are not independent, an interaction between the factors becomes an important part of the analysis.
4. Trial Timeline
Study start
The registered start date for INSPIRES is September 17, 2018.
Primary completion
The registry lists January 15, 2023 as the primary completion date.
The registry currently identifies the trial status as Unknown. The primary endpoint is evaluated at 90 days.
5. Interventions and Factorial Comparisons
| Treatment dimension | Level 1 | Level 2 |
|---|---|---|
| Antiplatelet therapy | Intensive antiplatelet | Standard antiplatelet |
| Statin timing | Immediate high-intensity statin | Delayed high-intensity statin |
Each treatment dimension has two registered levels, and the factorial design creates four study arms. The important statistical distinction is that the trial is not simply a four-group comparison with one treatment versus another. Its structure is intended to investigate two related treatment factors within one randomized experiment.
6. Primary Endpoint
| Endpoint | Time frame | Registry definition |
|---|---|---|
| Stroke | 90 days | Symptoms and signs of acute neurological deficits caused by sudden focal or whole brain, spinal cord or retinal vascular damage, which are related to cerebral circulatory disorders, including hemorrhagic and ischemic stroke |
The primary endpoint is therefore a clinical event assessed over a fixed 90-day period. Unlike a time-to-event endpoint such as overall survival, the registry definition here specifies a stroke outcome at a particular follow-up time rather than a duration until the event.
7. Planned Analysis
The registry identifies stroke at 90 days as the primary endpoint but does not post a formal statistical analysis for that endpoint. For a binary 90-day endpoint, the natural statistical representation is the proportion of participants experiencing stroke in each randomized treatment group or factorial treatment level.
A treatment comparison can then be expressed using an absolute risk difference, relative risk, odds ratio, or a regression model appropriate for a binary outcome. The appropriate prespecified analysis depends on the trial's detailed statistical analysis plan.
For the factorial design, a statistical model can represent the two treatment factors simultaneously. A basic formulation would include indicators for antiplatelet strategy and statin timing, with an interaction term if the design calls for formal assessment of whether the treatment effects depend on one another.
The interaction term addresses whether the effect associated with one treatment factor changes according to the level of the other factor.
8. Statistical Methodology
Binary outcome analysis
The 90-day stroke endpoint is naturally treated as a binary outcome: a participant either experiences the defined stroke endpoint within the specified 90-day period or does not. This differs from a continuous endpoint, where the magnitude of a measurement is directly analyzed, and from a time-to-event endpoint, where the exact timing of the event contributes information.
For a binary outcome, one straightforward summary is the event risk in each treatment group. The difference between two risks is an absolute risk difference, while their ratio is a relative risk.
These measures answer different questions. The risk difference describes the absolute change in event probability, whereas the relative risk describes the proportional relationship between the two risks.
Factorial analysis
The factorial design allows the effects of intensive versus standard antiplatelet therapy and immediate versus delayed high-intensity statin therapy to be evaluated within the same trial. This is statistically efficient when the treatment factors can reasonably be studied together.
The principal modeling issue is whether the two treatment factors act independently. If the effect of intensive antiplatelet therapy is approximately the same regardless of statin timing, an additive model without an interaction term may adequately describe the main effects. If the effect changes materially according to statin timing, an interaction becomes important for interpretation.
Interaction
An interaction means that the effect associated with one treatment factor depends on the level of the second factor. It is not established merely because the four observed group-specific event rates appear different.
Confidence intervals
For a treatment effect such as a risk difference or relative risk, a confidence interval describes the statistical uncertainty surrounding the estimated effect under the selected analysis model. Narrower intervals generally indicate greater precision, while wider intervals indicate more uncertainty.
P-values
A p-value addresses compatibility of the observed data with a specified null hypothesis under the statistical model. It does not measure the size of a treatment effect, the clinical importance of the effect, or the probability that the treatment hypothesis is true.
Randomization and causal interpretation
Randomization is central to the causal interpretation of a factorial trial. If treatment assignment is properly implemented, systematic differences in outcomes between randomized treatment strategies can be attributed to the assigned interventions under the assumptions of the randomized design, rather than to baseline differences selected by investigators or participants.
9. Statistical Methods Explained
Why use a factorial design?
A factorial design allows two treatment dimensions to be investigated within one randomized trial. INSPIRES has an antiplatelet factor and a statin-timing factor, each with two levels, producing four treatment arms. This can provide information about both treatment dimensions without requiring two entirely separate trials.
What is a main effect?
A main effect summarizes the average effect of one treatment factor across the levels of the other factor. For example, the antiplatelet main effect compares intensive versus standard antiplatelet treatment while averaging over the immediate and delayed statin strategies, subject to the assumptions of the factorial analysis.
What is an interaction?
An interaction occurs when the effect of one factor differs according to the level of the second factor. In INSPIRES, an interaction would mean that the relationship between intensive and standard antiplatelet therapy differs depending on whether high-intensity statin therapy is immediate or delayed.
Why is the 90-day endpoint analyzed as a binary outcome?
The registered primary endpoint specifies stroke during a 90-day time frame. A binary analysis asks whether the defined event occurred during that period. This makes the endpoint naturally interpretable as a proportion or risk, although the detailed statistical analysis plan determines the exact model and effect measure.
Why report absolute risk as well as relative effect?
A relative measure can describe proportional differences between treatment groups, but it does not show the absolute frequency of the event. Absolute risk differences put the treatment comparison on the probability scale and can therefore provide a complementary interpretation.
Why does randomization matter statistically?
Randomization separates treatment assignment from the prognostic characteristics of participants in expectation. This is what permits a difference in outcomes between randomized groups to be interpreted as a treatment effect, subject to adherence, follow-up, outcome ascertainment, and other assumptions relevant to the chosen estimand.
Why is the interaction term important in a factorial trial?
If the two interventions act independently, their effects can be summarized as main effects. If one treatment changes the effect of the other, a main-effect-only summary can be misleading. The interaction term provides a formal way to investigate this dependence.
10. Analysis Populations and Estimands
The registry identifies randomized allocation, a factorial treatment structure, and a 90-day primary endpoint. It does not provide a detailed definition of an intention-to-treat population, per-protocol population, safety population, or other analysis populations.
For a randomized treatment-effect analysis, the intention-to-treat principle generally retains participants according to their randomized assignment. This preserves the comparison created by randomization even when adherence or treatment exposure differs after assignment.
Randomized comparison
The fundamental comparison is based on treatment assignment rather than a post-randomization decision about which therapy a participant actually received.
Binary estimand
The primary outcome can be expressed as the probability of stroke by 90 days under the assigned treatment strategy.
Factorial estimand
Main effects summarize each treatment factor across the other factor when the factorial assumptions support that interpretation.
Interaction estimand
The interaction addresses whether the effect of one treatment factor changes according to the level of the other.
11. Masking and Outcome Assessment
The registry identifies INSPIRES as quadruple-masked. Masking can reduce the potential for knowledge of treatment assignment to influence participant behavior, clinical management, outcome assessment, or other aspects of trial conduct.
For a clinical endpoint such as stroke, masking is particularly relevant because outcome determination involves clinical assessment. The registry's endpoint definition specifies the neurological signs and symptoms included within the stroke endpoint and identifies both hemorrhagic and ischemic stroke within that definition.
12. Endpoint Definition in Statistical Context
The primary endpoint is not simply any neurological symptom. The registry defines stroke as symptoms and signs of acute neurological deficits caused by sudden focal or whole brain, spinal cord or retinal vascular damage related to cerebral circulatory disorders, including hemorrhagic and ischemic stroke.
| Element | Registered definition |
|---|---|
| Endpoint | Stroke |
| Time frame | 90 days |
| Clinical manifestation | Symptoms and signs of acute neurological deficits |
| Vascular scope | Sudden focal or whole brain, spinal cord or retinal vascular damage |
| Underlying context | Cerebral circulatory disorders |
| Stroke types explicitly included | Hemorrhagic and ischemic stroke |
Precise endpoint definitions are statistically important because the estimated event rate depends on which clinical events qualify as outcomes. A treatment comparison is only as interpretable as the consistency with which the endpoint is defined and assessed across randomized groups.
13. Enrollment and Precision
The registry reports an enrollment of 6100 participants across four study arms. A sample of this size provides a substantial amount of information for a binary primary endpoint, although the actual precision of the treatment-effect estimate depends on the number of observed stroke events, the distribution across factorial groups, and the analysis population.
For a binary endpoint, the number of participants alone does not determine precision. The event frequency and balance of information across treatment groups also affect the standard error and confidence interval.
Because the registry does not post a statistical analysis for the primary endpoint, an observed event count, treatment-effect estimate, confidence interval, or p-value cannot be used here to characterize the eventual precision of the INSPIRES result.
14. Multiplicity in a Factorial Trial
Factorial trials create an important multiplicity question because more than one treatment factor can generate a primary comparison. INSPIRES includes two treatment dimensions, each with two registered levels, and therefore naturally raises the distinction between the antiplatelet main effect, the statin-timing main effect, and a possible interaction.
| Statistical question | Meaning |
|---|---|
| Antiplatelet main effect | Comparison of intensive versus standard antiplatelet therapy across the statin-timing factor. |
| Statin-timing main effect | Comparison of immediate versus delayed high-intensity statin therapy across the antiplatelet factor. |
| Factorial interaction | Whether the effect of one treatment factor depends on the level of the other. |
| Primary endpoint | Stroke at 90 days. |
The appropriate multiplicity strategy depends on the prespecified statistical analysis plan. Multiple formal hypotheses can increase the probability of at least one false-positive finding if each is tested independently at the same nominal significance level. A factorial trial therefore benefits from an explicit hierarchy of primary questions and interaction testing strategy.
15. What a 90-Day Risk Estimate Means
If an analysis reported a 90-day stroke risk of R in a treatment group, that value would represent the estimated proportion of participants experiencing the registered stroke endpoint during the 90-day period under the specified analysis.
A risk difference compares the two absolute probabilities. A negative risk difference would indicate a lower estimated probability of stroke in the first group, while a positive value would indicate a higher estimated probability.
A relative risk compares the probabilities proportionally. A value below 1 would indicate a lower estimated event risk in the numerator group relative to the denominator group; it would not mean that the percentage value itself is an absolute reduction in stroke.
A confidence interval quantifies statistical uncertainty around the selected effect estimate. It does not describe the range of outcomes that individual participants will experience.
16. Safety
The registry information identifies the trial's four interventions and its primary stroke endpoint but does not provide posted serious-adverse-event results by arm.
Safety interpretation in a randomized factorial trial would ordinarily distinguish adverse events associated with each treatment factor from the primary efficacy outcome. Because the registered information does not post serious adverse-event counts or rates by arm, no arm-specific safety comparison is presented here.
17. Limitations
- No posted statistical analysis: the ClinicalTrials.gov record does not post a formal statistical analysis for the registered primary endpoint.
- No posted treatment-effect estimate: the registry information does not provide a stroke event count, risk estimate, effect estimate, confidence interval, or p-value for the 90-day primary endpoint.
- Factorial interaction: interpretation of the two treatment factors depends on whether their effects can reasonably be treated as independent or whether an interaction is present.
- Endpoint classification: the clinical definition of stroke determines which neurological events contribute to the primary analysis.
- Analysis-population detail: the registry information does not specify detailed intention-to-treat, per-protocol, or safety-population definitions.
- Multiplicity: factorial main effects and interaction testing create multiple statistical questions that require an appropriate prespecified error-control strategy.
- Precision: enrollment of 6100 participants does not by itself determine the precision of the primary treatment comparison; the number and distribution of observed stroke events are also important.
- Generalizability: the trial population is defined by acute stroke or transient ischemic attack and high-risk intracranial or extracranial atherosclerosis, so conclusions apply to the population represented by those eligibility criteria rather than automatically to every patient with cerebrovascular disease.
18. Why This Trial Matters Statistically
INSPIRES is a useful statistical teaching case because it combines randomized treatment allocation, quadruple masking, a factorial design, and a clinically defined fixed-time binary endpoint.
| Concept | How it appears in INSPIRES |
|---|---|
| Randomization | Participants are randomized to treatment strategies. |
| Factorial design | Four arms arise from two two-level treatment dimensions. |
| Multiple treatment factors | Antiplatelet intensity and statin timing are studied within one trial. |
| Interaction | The effect of one factor may depend on the level of the other. |
| Binary endpoint | Stroke is assessed over a specified 90-day period. |
| Risk difference | Provides an absolute comparison of stroke probabilities. |
| Relative risk | Provides a proportional comparison of stroke probabilities. |
| Confidence interval | Quantifies uncertainty around an estimated treatment effect. |
| Multiplicity | Multiple factorial hypotheses require an explicit statistical strategy. |
| Masking | Quadruple masking is part of the registered trial design. |
19. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The study is structured to compare two treatment dimensions within a randomized four-arm factorial design, using stroke at 90 days as the primary endpoint. The key statistical questions concern the two main effects and whether their effects interact.
Clinical interpretation
The clinical outcome is a defined stroke event occurring within 90 days among participants with acute stroke or transient ischemic attack and high-risk intracranial or extracranial atherosclerosis.
Keeping these perspectives separate is important. Statistical evidence describes the size, uncertainty, and structure of an observed treatment comparison. Clinical interpretation asks what that comparison means for the population and outcome being studied.
20. A Practical Analysis Framework
If the primary endpoint were analyzed from participant-level data, a clear statistical workflow would begin with the registered 90-day stroke definition and the randomized factorial assignment.
The final report should present both absolute and relative measures where appropriate. It should also state the analysis population, handling of missing 90-day outcomes, treatment of the factorial interaction, and the multiplicity strategy. These details are essential because the same underlying event data can produce different estimates or inferential conclusions under different estimands and models.
21. Missing Data and Follow-Up
The registered information identifies the primary endpoint and its 90-day time frame but does not specify a missing-data or imputation method.
For a fixed-time binary endpoint, missing 90-day outcome status can create uncertainty because the participant's stroke status is unknown at the analysis time point. The statistical analysis plan would ordinarily specify how such observations are handled and whether sensitivity analyses are performed under alternative assumptions.
22. What Would Make the Primary Result Informative?
A statistically informative primary analysis would combine several pieces of information rather than relying on a p-value alone.
| Component | Question answered |
|---|---|
| Event risk by 90 days | How frequently did the defined stroke endpoint occur? |
| Absolute risk difference | How large was the difference in event probability? |
| Relative risk or another relative measure | How did the event probability compare proportionally? |
| 95% confidence interval | How precisely was the treatment effect estimated? |
| P-value | How compatible are the data with the specified null hypothesis under the chosen model? |
| Interaction estimate | Does the effect of one factorial treatment depend on the other? |
This framework helps prevent a common statistical error: treating statistical significance as a measure of treatment magnitude. A small p-value can accompany a small effect, while a clinically meaningful effect can have substantial uncertainty if the available information is limited.
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Calculators
25. Sources
- ClinicalTrials.gov: NCT03635749 — INSPIRES.
Continue through Clinical Biostats
Explore statistical tutorials, calculators, and additional clinical trial analyses.
26. Record Summary
INSPIRES is a randomized phase 3 trial with a factorial design, four arms, quadruple masking, and an enrollment of 6100 participants. The registered interventions comprise intensive versus standard antiplatelet therapy and immediate versus delayed high-intensity statin therapy. Its primary endpoint is stroke at 90 days, defined by the registry as symptoms and signs of acute neurological deficits caused by sudden focal or whole brain, spinal cord or retinal vascular damage related to cerebral circulatory disorders, including hemorrhagic and ischemic stroke.
The central statistical feature is the factorial structure. The design permits evaluation of the two treatment dimensions within one randomized experiment and raises a specific question about interaction: whether the effect of one treatment factor depends on the level of the other. For the 90-day stroke endpoint, an appropriate analysis would ordinarily report absolute event risks together with an effect measure such as a risk difference or relative risk, corresponding confidence intervals, and an inferential assessment under the prespecified statistical plan.