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
TARDIS was a phase 3 randomized parallel-group clinical trial evaluating triple antiplatelet therapy after ischaemic stroke.
| Feature | TARDIS |
|---|---|
| Phase | Phase 3 |
| Status | Terminated |
| Start date | 2009-04 |
| Primary completion date | 2017-09 |
| Lead sponsor | University of Nottingham |
| Condition | Stroke |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Single |
| Primary purpose | Treatment |
2. Clinical Question
The clinical question addressed by TARDIS was whether treatment involving aspirin, dipyridamole, and clopidogrel could be evaluated against the comparator strategy in patients with stroke using a randomized clinical trial framework.
Population
Patients with stroke.
Intervention
Aspirin, dipyridamole, and clopidogrel.
Comparator
The randomized comparator arm.
Primary question
Comparison of ordinal stroke severity at 90 days.
3. Trial Design
3096 patients
2 arms
Phase 3
90 days
Aspirin, dipyridamole, and clopidogrel.
Randomized parallel-group treatment study.
4. Endpoints
| Endpoint | Definition | Time frame |
|---|---|---|
| Primary outcome | Ordinal stroke severity at 90 days using a 5-level ordinal stroke and TIA scale ordered by severity using the modified Rankin Scale (mRS): fatal stroke / severe non-fatal stroke (mRS 2-5) / mild stroke (mRS 0,1) / TIA / no stroke-TIA. | 90 days |
5. Planned Analysis
The registry identifies the primary outcome as an ordinal stroke severity measure rather than a simple binary event. Ordinal outcomes preserve information about severity categories and can provide greater statistical efficiency than collapsing outcomes into only stroke versus no stroke.
Ordinal outcome analysis
For an ordered categorical endpoint, statistical approaches commonly model the probability distribution across severity levels while respecting the ordering between categories.
The registry describes the outcome scale as an approach that allows smaller sample sizes compared with binary outcomes such as stroke/no stroke.
6. Statistical Methodology
Ordinal analysis
The primary endpoint uses ordered categories of stroke severity. Unlike a binary endpoint, an ordinal analysis uses the full ranking of outcomes, distinguishing between different degrees of disability and clinical status.
Why ordinal outcomes matter
Converting an ordinal scale into a binary outcome can discard information. For example, severe disability, mild disability, and no event are clinically different states. Ordinal methods attempt to retain these distinctions.
Randomization
Random allocation is designed to balance measured and unmeasured factors between treatment groups, allowing differences in outcomes to be interpreted within the randomized comparison.
7. Statistical Methods Explained
Why use an ordinal stroke severity outcome?
An ordinal endpoint captures the degree of neurological outcome rather than only whether a stroke occurred. This can increase statistical efficiency when clinically meaningful severity categories are available.
How is an ordinal endpoint different from a binary endpoint?
A binary endpoint has two possible outcomes. An ordinal endpoint contains ordered categories, allowing analysis of the distribution across levels of severity.
What does the modified Rankin Scale contribute?
The modified Rankin Scale provides an ordered framework for describing disability after stroke, allowing outcomes to be grouped by increasing severity.
Why does randomization matter?
Randomization creates comparable groups at baseline on average, reducing confounding in the treatment comparison.
Why are confidence intervals important?
Confidence intervals describe uncertainty around estimated effects. They provide information about the precision of an estimate rather than the probability that a particular value is true.
8. Limitations
- No posted results: The ClinicalTrials.gov record does not report outcome estimates or formal statistical analyses.
- Endpoint interpretation: Ordinal outcomes require assumptions about how severity categories relate statistically.
- Generalizability: Applicability depends on the enrolled stroke population and study design.
- Missing information: The registry does not report additional analysis details such as subgroup analyses, missing-data methods, interim analyses, or multiplicity adjustments.
9. Why This Trial Matters Statistically
TARDIS is a useful teaching example because it illustrates how clinical trials can use ordinal endpoints to preserve clinically meaningful information rather than reducing outcomes to a simple yes/no event.
| Concept | How it appears in TARDIS |
|---|---|
| Randomization | Randomized phase 3 parallel-group design. |
| Ordinal outcomes | Primary outcome based on ordered stroke severity categories. |
| Modified Rankin Scale | Framework for ordered disability severity assessment. |
| Clinical interpretation | Separating statistical modeling from clinical meaning of severity categories. |
10. Sources
- ClinicalTrials.gov record: TARDIS (NCT01661322)