This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. The ClinicalTrials.gov record indicates that results have been posted, but no formal statistical analyses are included in the registry-reported statistical-analysis field.
1. Trial at a Glance
SWIFT PRIME was a randomized, parallel-group clinical trial in acute ischemic stroke comparing intravenous recombinant human tissue plasminogen activator (rtPA) with rtPA plus the Solitaire revascularization device. The registered primary endpoint was 90-day global disability assessed through blinded evaluation of the modified Rankin Scale (mRS).
| Feature | SWIFT PRIME |
|---|---|
| Trial name | Solitaire™ With the Intention For Thrombectomy as PRIMary Endovascular Treatment (SWIFT PRIME) Trial |
| Clinical condition | Acute Ischemic Stroke |
| Therapeutic area | Neurology |
| Design | Randomized, parallel-group |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 196 |
| Arms | 2 |
| Status | Completed |
| Lead sponsor | Medtronic Neurovascular Clinical Affairs |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT01657461 |
2. Clinical Question
The trial addresses a treatment question in acute ischemic stroke: among the randomized study population, does adding the Solitaire revascularization device to intravenous rtPA change the degree of disability assessed 90 days after treatment compared with intravenous rtPA alone?
Population
Participants enrolled in the SWIFT PRIME trial with the registered condition of acute ischemic stroke.
Intervention
Intravenous recombinant human tissue plasminogen activator (rtPA) together with the Solitaire revascularization device.
Comparator
Intravenous rtPA.
Primary question
How does randomized treatment assignment relate to global disability at 90 days as assessed by the blinded modified Rankin Scale evaluation?
3. Trial Design
The ClinicalTrials.gov record describes SWIFT PRIME as a randomized, parallel-group trial with no masking and a primary purpose of treatment. The overall enrollment was 196 participants, with 2 study arms.
IV rtPA + Solitaire
- Intravenous recombinant human tissue plasminogen activator (rtPA)
- Solitaire revascularization device
- Serious adverse events: 35/98 in the ClinicalTrials.gov record
IV rtPA
- Intravenous recombinant human tissue plasminogen activator (rtPA)
- No additional device intervention is listed for this comparator
- Serious adverse events: 30/97 in the ClinicalTrials.gov record
4. Endpoints
The ClinicalTrials.gov record identifies 1 registered primary endpoint. Results are marked as posted, but the registry-reported statistical-analysis field contains no formal statistical comparison.
| Endpoint | Time frame | Registry description |
|---|---|---|
| 90-day Global Disability Assessed Via the Blinded Evaluation of Modified Rankin Score (mRS). | 90 days | mRS is a scale for measuring the degree of disability or dependence in the daily activities of people who have suffered a stroke: 0 No symptoms at all; 1 No significant disability despite symptoms, able to carry out all usual duties and activities; 2 Slight disability, unable to carry out all previous activities, but able to look after own affairs without assistance; 3 Moderate disability, requiring ... |
Why the primary endpoint is clinically and statistically different from a binary outcome
The modified Rankin Scale is an ordinal disability scale rather than a simple continuous measurement. Its categories represent increasing levels of functional dependence. That structure matters statistically because a comparison can preserve more information when it evaluates the ordered distribution of disability rather than reducing every participant to a single binary threshold.
The registered endpoint also specifies blinded evaluation. Blinding the outcome assessment is particularly relevant for a functional disability endpoint because the treatment itself and the absence of masking in the trial design could otherwise make assessment more vulnerable to observer expectations.
5. Planned Analysis
The ClinicalTrials.gov record contains posted outcome measures, including the 90-day mRS endpoint, but its no formal statistical analyses were posted. Therefore, no formal treatment-effect estimate, confidence interval, or p-value can be reported from the ClinicalTrials.gov record.
For an ordinal outcome such as mRS, an ordinal or proportional-odds analysis is one common statistical framework. A binary responder analysis at a prespecified mRS threshold is another possible framework. The appropriate method depends on the prespecified statistical analysis plan and the exact endpoint definition.
Why the statistical method cannot be assigned retrospectively
Knowing that the endpoint is an ordinal disability scale does not establish which formal analysis the SWIFT PRIME investigators prespecified. Several methods can be reasonable for ordinal outcomes, but they answer subtly different questions and require different assumptions. For example, a proportional-odds model uses the ordering of mRS categories and summarizes the odds of being at or below successive thresholds under a common odds-ratio assumption.
Likewise, a dichotomized mRS endpoint would answer a different question: whether the proportion of participants achieving a prespecified level of functional outcome differs between groups. Dichotomization can be useful when a threshold has a clear clinical interpretation, but it discards some of the ordering information contained in the original scale.
The central statistical issue is therefore not simply whether mRS was measured. It is how the full ordinal distribution was translated into an estimand. Without the formal statistical analysis field or an analysis plan in the ClinicalTrials.gov record, selecting one method and presenting its result as the SWIFT PRIME analysis would go beyond the evidence provided here.
6. Statistical Methods Explained
Why is randomization important for the primary comparison?
Randomization creates the basis for comparing outcomes between treatment assignments without relying solely on observed baseline characteristics to construct comparable groups. In a randomized trial, the treatment assignment is determined by the allocation process rather than by the participant's subsequent outcome.
This does not guarantee identical groups in every measured or unmeasured characteristic. Rather, randomization provides the design framework that allows a treatment-group comparison to be interpreted causally under the trial's assumptions.
Why is the modified Rankin Scale an ordinal endpoint?
The mRS categories have a meaningful order from less disability to greater disability, but the numerical labels should not automatically be treated as equally spaced measurements. The difference between adjacent categories is not necessarily the same clinical distance at every point on the scale.
That is why simply calculating an arithmetic mean of mRS values and applying a standard continuous-outcome model can impose assumptions that need explicit justification. Ordinal methods instead focus on the ordered nature of the categories.
What does a proportional-odds model mean for mRS?
An ordinal proportional-odds model can be expressed conceptually as a common odds ratio across the cumulative mRS thresholds. For example, the model considers the odds of being at or below a given disability category and uses the ordering of all categories together.
Here, Y is the mRS outcome, k indexes the ordered thresholds, and X represents treatment assignment. The proportional-odds assumption means the treatment coefficient is common across the cumulative thresholds.
A formal analysis of SWIFT PRIME would require the actual prespecified model and its assumptions before an estimated odds ratio could be attributed to the trial.
Why does blinded endpoint assessment matter?
The registry specifically describes the primary endpoint as a blinded evaluation of mRS. This separates the measurement of disability from knowledge of treatment assignment at the assessment stage. That is particularly valuable when the endpoint requires clinical judgment rather than a laboratory instrument producing an automatically recorded number.
Why should the 90-day time frame be preserved?
The registered endpoint is explicitly defined at 90 days. A time-specific functional endpoint is different from a lifetime outcome or an unspecified follow-up measurement. Changing the time frame would change the estimand and could make the resulting analysis incomparable with the registered endpoint.
Why can the safety denominator differ from enrollment?
Safety analyses often use an exposure-defined population rather than everyone originally enrolled. In the ClinicalTrials.gov record, serious adverse events are reported as 35/98 and 30/97. Because these denominators differ from the overall enrollment of 196, the safety figures should be presented with their registry-reported denominators rather than replacing them with the enrollment total.
7. Results: What the ClinicalTrials.gov record Actually Report
ClinicalTrials.gov marks the trial as having results posted and identifies 10 posted outcome measures. However, no formal statistical analyses were posted. Consequently, the ClinicalTrials.gov record does not provide a formal statistical comparison for the primary 90-day mRS endpoint.
| Evidence component | What is available in the ClinicalTrials.gov record |
|---|---|
| Primary endpoint | 90-day global disability assessed via blinded evaluation of mRS |
| Results posted | Yes |
| Outcome measures posted | 10 |
| Formal statistical analyses reported | None |
| Primary-effect estimate reported | None |
| Confidence interval reported | None |
| p-value reported | None |
8. Safety Results
The ClinicalTrials.gov record provides serious adverse events by arm. These are the only safety results available in the ClinicalTrials.gov recordset and are therefore reported without adding other adverse-event categories or percentages.
| Study arm | Serious adverse events | Affected / at risk |
|---|---|---|
| IV t-PA With Solitaire™ Revascularization | Serious adverse events | 35/98 |
| IV t-PA | Serious adverse events | 30/97 |
The counts alone should not be converted into a comparative safety conclusion without preserving the registry-reported denominators and knowing the prespecified safety-analysis population and follow-up rules. The denominators are 98 and 97, respectively, rather than the overall enrollment of 196.
The ClinicalTrials.gov record indicates 35 affected participants among 98 at risk in the IV t-PA plus Solitaire arm and 30 among 97 at risk in the IV t-PA arm. Those figures describe the reported serious-adverse-event counts and denominators; they do not by themselves provide a confidence interval, hypothesis test, adjusted risk ratio, or causal safety estimate.
9. Statistical Methodology
Randomization as the primary design principle
SWIFT PRIME is explicitly identified as randomized. Randomization is the foundation for comparing the intervention and comparator groups because it creates treatment groups through an allocation mechanism rather than through clinical selection.
For the primary endpoint, the key analysis principle is to preserve the randomized comparison. If participants are excluded selectively after randomization based on post-randomization information, the resulting comparison can lose some of the protection provided by the original design.
Ordinal outcome analysis
The 90-day mRS is ordered. A statistically efficient analysis can use that ordering rather than treating the six-level or multi-category scale as nominal categories. One common approach is proportional-odds regression, although the registry-reported SWIFT PRIME data do not identify that method as the actual prespecified analysis.
The exact direction and interpretation depend on how the cumulative outcome is coded. A formal trial result should therefore always state the coding convention alongside the estimated effect.
Confidence intervals
A confidence interval would quantify statistical uncertainty around the estimated treatment effect under the selected model. It would not describe the range of individual patient outcomes, nor would it tell us the probability that the true effect lies inside the particular interval after the data have been observed.
P-values
A p-value, if reported, would quantify the compatibility of the observed data with a specified null hypothesis under the chosen statistical model. It would not measure the magnitude or clinical importance of the treatment effect. An effect estimate and its confidence interval are therefore essential companions to a p-value.
Multiplicity
The ClinicalTrials.gov record identifies 1 registered primary endpoint but does not provide an alpha-allocation scheme, hierarchy of secondary endpoints, multiplicity adjustment, or interim-testing plan. Those features should not be inferred from the fact that 10 outcome measures were posted.
Missing data
The ClinicalTrials.gov record does not describe an imputation strategy, treatment of participants without a 90-day mRS assessment, or sensitivity analyses for missing functional outcomes. These details can materially affect an ordinal endpoint and should be taken from the prespecified analysis plan when available rather than reconstructed from the posted endpoint alone.
10. Crossover, Interim Analysis, and Other Design Topics
The ClinicalTrials.gov record does not provide evidence about crossover rules, a factorial design, a non-inferiority margin, an interim-analysis boundary, a Bayesian analysis, or a specific missing-data/imputation method.
| Design topic | Evidence in the ClinicalTrials.gov record | Statistical implication |
|---|---|---|
| Non-inferiority margin | Not provided | No non-inferiority interpretation can be made. |
| Crossover | Not provided | No crossover adjustment or dilution analysis can be described. |
| Factorial design | Not provided | The registry identifies a parallel design, not a factorial structure. |
| Multiplicity | Not provided | No formal multiplicity strategy should be inferred. |
| Interim analysis | Not provided | No alpha-spending or stopping-boundary analysis can be reported. |
| Missing data / imputation | Not provided | No imputation method should be attributed to the trial. |
| Stratification | Not provided | No stratified analysis factors are reported. |
| Bayesian methods | Not provided | No Bayesian model or posterior analysis can be reported. |
This distinction is important. A statistical analysis page should explain methods that are relevant to the endpoint while separating those methods from methods that the registry actually documents as having been used.
11. Interpreting an Ordinal Functional Endpoint
The mRS illustrates why clinical-trial statistics cannot be reduced to a single p-value. A participant's score represents a level of functional disability, and the categories have an intrinsic order. The scientific question can therefore be framed in several ways, depending on the prespecified estimand.
Distributional question
Does the overall distribution of functional disability shift toward more favorable mRS categories under one randomized treatment assignment?
Threshold question
Does treatment change the probability of achieving a prespecified functional-outcome threshold?
Continuous-score question
Can the ordered categories reasonably be represented numerically for a mean-score analysis? This requires assumptions that should be explicitly justified.
Clinical interpretation
Which changes in disability correspond to differences that matter to patients and clinicians? Statistical significance alone does not answer this question.
These are not interchangeable estimands. A page that reports an odds ratio from an ordinal model, a risk ratio from a binary threshold, and a mean difference as though they measured the same treatment effect would obscure the actual statistical question.
12. Blinded Evaluation and Measurement Error
The registered primary endpoint explicitly uses a blinded evaluation of the modified Rankin Scale. Blinding is a measurement safeguard rather than a substitute for randomization.
Randomization addresses allocation. Blinded outcome assessment addresses the possibility that knowledge of treatment assignment could influence the recorded outcome. For a functional scale that requires clinical assessment, keeping these roles separate is particularly important when interpreting the credibility of an observed between-group difference.
Even a perfectly randomized trial can be affected by measurement error. If outcome assessment differs systematically according to treatment knowledge, the measured treatment effect can be distorted. Blinded assessment is therefore part of the design supporting the validity of the primary endpoint measurement.
13. Trial Timeline
Trial start
The ClinicalTrials.gov record lists December 2012 as the study start.
Primary completion
The ClinicalTrials.gov record lists January 2015 as the primary completion date.
Registry status
The study status is listed as completed, with results posted and 10 outcome measures posted in the ClinicalTrials.gov record.
14. What a Formal Primary Analysis Would Need to Report
Because the registry-reported statistical-analysis field is empty, the most useful statistical framework is to identify what a complete report would need to establish before an effect estimate could be interpreted.
| Component | Why it matters |
|---|---|
| Analysis population | Defines which randomized participants contribute to the primary comparison. |
| mRS coding | Determines the direction of a favorable outcome and the exact estimand. |
| Statistical model | Determines how the ordered categories are translated into an effect estimate. |
| Model assumptions | Determines whether the selected statistical interpretation is justified. |
| Effect estimate | Quantifies the magnitude and direction of the treatment comparison. |
| Confidence interval | Describes statistical uncertainty around the estimate. |
| Hypothesis test | Provides the prespecified inferential comparison against the null hypothesis. |
| Missing-data handling | Explains how participants without evaluable 90-day mRS assessments were handled. |
| Multiplicity strategy | Determines how multiple confirmatory comparisons affect type I error. |
This framework illustrates why a result cannot be responsibly reconstructed from the endpoint name alone. The statistical estimand is defined jointly by the outcome, coding, analysis population, model, assumptions, missing-data rules, and inferential framework.
15. Serious Adverse Events: Statistical Interpretation
The serious-adverse-event data provide affected participants and denominators for each arm. The registry-reported figures are 35/98 for IV t-PA with Solitaire and 30/97 for IV t-PA.
These are descriptive safety counts. The ClinicalTrials.gov recordset does not provide a formal statistical comparison of serious adverse-event rates.
Several analytical decisions would be needed before turning these data into an inferential safety estimate: the precise definition and observation window for serious adverse events, the population used for the denominator, handling of repeated events, and whether the prespecified analysis compares participants with at least one event or total event counts.
For example, a comparison of the proportion of participants with at least one serious adverse event is not the same as a comparison of the total number of serious adverse events. Repeated events can make the latter a count-data problem rather than a simple binary outcome.
16. Limitations
- No formal statistical analyses reported: no formal statistical analyses were posted, so treatment-effect estimates, confidence intervals, and p-values cannot be reported from the ClinicalTrials.gov record.
- Limited safety detail: the ClinicalTrials.gov record contains serious-adverse-event counts and denominators but no broader safety table or formal comparison.
- Analysis population not specified: the ClinicalTrials.gov record does not identify whether the primary analysis uses all randomized participants, an evaluable population, or another prespecified population.
- Missing-data strategy not specified: no imputation or sensitivity-analysis method is posted on ClinicalTrials.gov for participants without a 90-day mRS assessment.
- Multiplicity not specified: the registry record identifies 1 primary endpoint but does not provide an alpha level, adjustment strategy, or hierarchy for the other posted outcomes.
- Stratification not specified: no stratification factors are provided in the ClinicalTrials.gov record.
- Model assumptions not documented: no specific ordinal regression, binary regression, or other primary statistical model is identified in the registry-reported statistical-analysis field.
- Safety denominators differ from enrollment: the serious-adverse-event denominators are 98 and 97, so they should not be silently replaced with the overall enrollment of 196.
17. Why This Trial Matters Statistically
SWIFT PRIME is a useful teaching case because it connects randomized treatment allocation with an important patient-centered ordinal outcome. The statistical challenge is not simply to determine whether two numbers differ; it is to preserve the information contained in the ordered disability scale while maintaining the causal structure created by randomization.
| Concept | How it appears in SWIFT PRIME |
|---|---|
| Randomization | The trial is explicitly described as randomized. |
| Parallel-group design | The registry identifies a parallel design with 2 arms. |
| Ordinal outcome | The primary endpoint uses the modified Rankin Scale. |
| Blinded assessment | The primary endpoint is based on blinded evaluation of mRS. |
| Time-specific endpoint | The primary endpoint is assessed at 90 days. |
| Analysis population | Must be defined before interpreting an effect estimate; not reported here. |
| Missing data | Handling of participants without evaluable 90-day outcomes is not reported. |
| Multiplicity | One registered primary endpoint is identified; the formal multiplicity strategy is not reported. |
| Safety denominators | Serious adverse events are reported as 35/98 and 30/97. |
The most important statistical lesson is that endpoint definition and statistical estimand are inseparable. The same mRS data can support different statistical summaries, but each summary answers a different question. A rigorous trial analysis therefore reports the prespecified method rather than selecting a method after seeing the outcome.
18. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The trial randomized 196 participants across 2 parallel arms and specified a 90-day blinded mRS assessment as its primary endpoint. The ClinicalTrials.gov record does not provide the formal statistical comparison for that endpoint.
Clinical interpretation
The clinical outcome of interest is global disability after acute ischemic stroke. Because the ClinicalTrials.gov record does not contain the comparative mRS results, the clinical magnitude of any treatment difference cannot be stated from the registry record.
Keeping these statements separate prevents an important analytical error: treating the existence of a randomized trial and a posted outcome as though they automatically establish the magnitude, precision, or statistical significance of an observed treatment effect.
19. Related Tutorials
Learn more about the methods used in this trial:
20. Related Statistical Calculators
21. Sources
- ClinicalTrials.gov: SWIFT PRIME, NCT01657461.
- Linked publication: PubMed record — PMID 31239326.
- Linked publication: PubMed record — PMID 30210431.
- Linked publication: PubMed record — PMID 30045949.
- Linked publication: PubMed record — PMID 29135347.
- Linked publication: PubMed record — PMID 29127195.
Continue through the Clinical Biostats statistical tutorials
Explore the statistical concepts that connect randomized trial design, ordinal endpoints, clinical outcome assessment, and inferential analysis.
22. Record Summary
SWIFT PRIME is a randomized, parallel-group trial with 196 enrolled participants and 2 arms, evaluating intravenous rtPA with or without the Solitaire revascularization device in acute ischemic stroke. Its registered primary endpoint is global disability at 90 days assessed through blinded evaluation of the modified Rankin Scale.
From a statistical perspective, the key feature is the ordinal nature of the primary endpoint. An appropriate analysis must account for the ordering of mRS categories and clearly define the estimand, analysis population, model, assumptions, missing-data strategy, and inferential framework. The ClinicalTrials.gov record identifies the endpoint and indicate that results were posted, but they do not provide a formal statistical-analysis record. The responsible interpretation is therefore to distinguish the documented trial design and endpoint from any treatment-effect estimate that would require additional source data.