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
RESCUE-Japan LIMIT was a randomized, parallel-group clinical trial evaluating endovascular treatment for acute large vessel occlusion with large ischemic core. The registry lists 203 enrolled participants and a primary endpoint based on functional outcome at 90 days after stroke.
| Feature | RESCUE-Japan LIMIT |
|---|---|
| Study title | Randomized Controlled Trial of Endovascular Therapy for Acute Large Vessel Occlusion With Large Ischemic Core |
| ClinicalTrials.gov identifier | NCT03702413 |
| Status | Completed |
| Phase | NA |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 203 |
| Number of arms | 2 |
| Lead sponsor | Hyogo Medical University |
| Sponsor type | Other |
| Condition | Mechanical Thrombectomy; Large Ischemic Core |
| Intervention | Endovascular treatment (procedure) |
2. Clinical Question
The clinical question is whether endovascular treatment should be evaluated for patients with acute large vessel occlusion and large ischemic core, using functional outcome at 90 days as the primary endpoint.
Population
Patients represented by the registry conditions of mechanical thrombectomy and large ischemic core in the setting of acute large vessel occlusion.
Intervention
Endovascular treatment, recorded in the registry as a procedure.
Comparator
The study has 2 arms, but the registry information presented here does not specify the intervention assigned to the other arm.
Primary question
How does randomized assignment relate to achieving a modified Rankin Scale score of 3 or less at 90 days after stroke?
3. Trial Design
The trial began on 2018-11-30 and reached primary completion on 2021-12-21. The registry lists 203 enrolled participants across 2 arms.
Why randomization is important here
For a treatment-effect question, randomization is the key design feature that allows differences in outcomes between the assigned groups to be interpreted as evidence about the treatment comparison, subject to the usual assumptions of a randomized trial. Randomization does not guarantee identical groups on every characteristic; rather, it provides the probability-based mechanism that supports comparability in expectation.
4. Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Modified Rankin Scale ≤3 at 90 days | The primary endpoint is the modified Rankin Scale (mRS) ≤3 at 90 days post-stroke. The scale runs from 0-6, with 0 being perfect health without symptoms to 6 being death. | 90 days |
Understanding the modified Rankin Scale
The modified Rankin Scale is an ordered functional-outcome scale. The registry defines the levels as follows:
| mRS | Registry description |
|---|---|
| 0 | No symptoms. |
| 1 | No significant disability. Able to carry out all usual activities, despite some symptoms. |
| 2 | Slight disability. Able to look after own affairs without assistance, but unable to carry out all previous activities. |
| 3 | Moderate disability. Requires some help, but able to walk unassisted. |
| 4 | Moderately severe disability. Unable to attend to own bodily needs without assistance, and unable to walk unassisted. |
| 5 | Severe disability. Requires constant nursing care and attention, bedridden, incontinent. |
| 6 | Dead. |
Why the primary endpoint is dichotomized
The registry defines the primary endpoint as mRS ≤3 rather than as the complete 0-to-6 distribution. This converts the functional scale into a binary outcome: participants either meet the threshold of 3 or less at 90 days, or they do not.
That distinction matters statistically. The original mRS contains more information than a binary threshold because it preserves distinctions among scores from 0 through 6. A binary endpoint is easier to summarize and compare directly, but it discards some of that ordering information. The choice of the threshold therefore determines the precise clinical question being tested.
5. Statistical Methodology
No formal statistical analyses are posted in the ClinicalTrials.gov record. The registry does, however, identify a clearly defined binary primary endpoint, which allows the statistical structure of the intended comparison to be explained.
Binary endpoint analysis
The primary endpoint can be represented as an indicator variable:
The proportion with Y = 1 is the event rate for the favorable functional-outcome threshold.
For a randomized comparison with a binary primary endpoint, a conventional analysis would compare the proportion achieving the endpoint between the treatment and comparator arms. Common effect measures include the risk difference, risk ratio, and odds ratio.
Risk difference
The risk difference compares the absolute probability of achieving mRS ≤3 between the 2 randomized groups:
A positive risk difference would indicate a higher observed proportion achieving mRS ≤3 in the treatment group.
The risk difference is particularly intuitive because it remains on the probability scale. For example, an absolute difference of 0.10 would correspond to a 10-percentage-point difference. No such observed difference is reported in the registry record.
Risk ratio
The risk ratio compares the probability of achieving the endpoint in one arm with the probability in the other:
A risk ratio above 1 would indicate that the probability of achieving mRS ≤3 is higher in the treatment group.
Odds ratio
A logistic-regression or 2-by-2-table analysis may express the treatment effect as an odds ratio:
The odds ratio compares odds, not probabilities. When the endpoint is common, the odds ratio can differ substantially from the risk ratio.
Confidence intervals
An appropriate analysis would normally accompany the treatment-effect estimate with a confidence interval. The interval describes statistical uncertainty around the estimated effect under the chosen model and sampling framework. It does not describe the range of outcomes that an individual patient might experience.
Hypothesis testing
A binary endpoint can also be tested using a contingency-table method such as a chi-square test or Fisher's exact test, depending on the analysis plan and cell counts. Regression-based approaches can additionally adjust for prespecified covariates or account for the randomized design. The registry information available here does not identify which of these methods was prespecified.
6. Planned Analysis
The primary endpoint is modified Rankin Scale ≤3 at 90 days. Because this is a binary functional-outcome endpoint, the principal statistical comparison would typically estimate the proportion of participants achieving mRS ≤3 in each randomized arm and compare those proportions using an appropriate binary-outcome method.
Primary estimand
The treatment comparison concerns the probability of achieving mRS ≤3 at 90 days under the randomized treatment assignments.
Effect measures
Risk difference, risk ratio, and odds ratio are standard ways to quantify a difference in binary outcomes. The appropriate prespecified measure should be used for formal reporting.
Uncertainty
A confidence interval should accompany the primary treatment-effect estimate so that the precision of the comparison is visible rather than relying on a p-value alone.
Primary time point
The registry specifies 90 days post-stroke as the time frame for the primary endpoint.
7. Statistical Methods Explained
Why is the primary endpoint binary?
The registry defines success as an mRS score of 3 or less at 90 days. This produces a clinically interpretable threshold: the analysis asks how often participants reach that specified functional-outcome level. The trade-off is that the analysis no longer distinguishes among the individual scores within the favorable and unfavorable categories.
What does a risk difference mean?
A risk difference is the subtraction of one group's event probability from the other's. In this trial, it would quantify the absolute difference in the probability of achieving mRS ≤3 at 90 days. Unlike a relative measure, it directly expresses the size of the difference in percentage-point terms.
What does a risk ratio mean?
A risk ratio expresses one group's probability of achieving mRS ≤3 relative to the other group's probability. A value of 1 would indicate equal probabilities. Values above 1 would indicate a higher probability in the numerator group.
Why is an odds ratio different from a risk ratio?
An odds ratio operates on odds rather than probabilities. Odds are calculated as p/(1−p). When the outcome is frequent, the numerical difference between an odds ratio and a risk ratio can be substantial. Therefore, an odds ratio should not automatically be described as a relative increase in probability.
Why does the 90-day time frame matter?
The endpoint is explicitly tied to 90 days after stroke. The same functional scale measured at a different time would represent a different endpoint because recovery after stroke changes over time. The analysis therefore needs to preserve the registry-defined 90-day assessment window.
Why might the complete mRS scale be analyzed differently?
The mRS is inherently ordered from 0 through 6. An analysis that retains the complete ordinal scale can potentially use more information than the dichotomized mRS ≤3 endpoint. Ordinal methods, however, answer a different statistical question from the registered binary endpoint and should not be substituted silently for the prespecified primary outcome.
8. Interpreting the Primary Endpoint Correctly
The primary endpoint does not ask whether a participant improved by a particular number of mRS points. It asks whether the participant's score is 3 or less at 90 days. A participant with an mRS of 0 and a participant with an mRS of 3 are therefore both counted as having achieved the primary endpoint.
If an analysis reports both a risk difference and a risk ratio, they should be read together. The risk difference describes the absolute separation between groups, while the risk ratio describes the relative probability. Neither measure alone completely describes the distribution of functional outcomes across the full mRS scale.
A p-value addresses the compatibility of the observed data with a specified null hypothesis under the statistical model. It does not quantify how large or clinically important the treatment difference is. The estimated effect and its confidence interval are needed to understand magnitude and precision.
The binary threshold is the registered primary endpoint, but it compresses a 7-level functional scale into 2 categories. Two treatments could have the same proportion of participants at mRS ≤3 while producing different distributions within the individual mRS categories. The binary endpoint therefore answers a focused question rather than describing every aspect of functional recovery.
9. Analysis Population and Randomization
The registry identifies the allocation as randomized and the design as parallel. For a randomized treatment comparison, an intention-to-treat framework would ordinarily preserve the benefits of randomization by analyzing participants according to their assigned treatment arm.
| Design feature | Statistical implication |
|---|---|
| Randomized allocation | Supports comparison of outcome probabilities between assigned treatment groups. |
| Parallel design | Each participant contributes to one randomized study arm rather than serving sequentially in multiple arms. |
| No masking | The registry describes the study as unmasked, which can be relevant when outcomes or treatment decisions are susceptible to knowledge of assignment. |
| 203 enrolled | The total randomized study population listed in the registry is 203. |
For the primary efficacy question, the analysis population should be defined before examining outcomes. Excluding participants after randomization based on post-randomization characteristics can compromise the comparability created by randomization. The registry information presented here does not specify the detailed analysis-population rules.
10. Missing Data and Endpoint Assessment
The primary endpoint requires a modified Rankin Scale assessment at 90 days. A participant without an evaluable 90-day outcome creates a missing binary endpoint and therefore raises an important analysis issue.
In a formal statistical analysis, the treatment of missing 90-day assessments should be prespecified. Possible approaches include defining an analysis population based on evaluability, using an explicitly specified missing-data rule, or applying a model-based approach under stated assumptions. Each choice can affect the estimated treatment comparison.
The registry information presented here does not specify the missing-data or imputation method. That distinction is important because the method used for participants without an observed 90-day mRS can influence both the estimate and its uncertainty.
11. Why the Endpoint Choice Is Statistically Important
The primary endpoint combines a clinically ordered scale with a prespecified threshold. This creates a useful example of how endpoint definition determines the statistical problem.
Preserves a clinical threshold
mRS ≤3 identifies a defined level of functional outcome rather than asking only whether the score changed.
Creates a binary outcome
The statistical comparison becomes a comparison of proportions achieving the specified threshold.
Discards some ordering information
Scores within the favorable and unfavorable categories are treated identically for the primary binary analysis.
Fixes the assessment time
The registered endpoint is evaluated at 90 days post-stroke, making timing part of the endpoint definition.
12. Interpreting the Modified Rankin Scale as an Ordered Outcome
The mRS contains meaningful ordering: an mRS of 2 represents less disability than an mRS of 4, while 6 represents death. A dichotomized endpoint retains only one threshold from that ordered structure.
This creates two distinct analytical perspectives. The registered primary endpoint asks whether the patient reaches the threshold of 3 or less. An ordinal analysis would instead ask whether the treatment shifts the distribution of mRS scores toward better outcomes across the full scale. These are complementary questions, but they are not mathematically interchangeable.
The indicator function converts the ordered scale into a binary endpoint for the registered primary analysis.
This distinction is particularly important when interpreting a reported proportion. A higher proportion with mRS ≤3 establishes a difference at the selected threshold; it does not by itself establish how the treatment changed the relative frequencies of mRS 0, 1, 2, 3, 4, 5, and 6.
13. Statistical Precision and Sample Size
The registry lists 203 enrolled participants. The precision of a binary treatment comparison depends not only on the total sample size but also on the number of participants in each arm, the observed endpoint probabilities, and the analysis method.
For a proportion, sampling variability is related to both the event probability and the number of observations. Consequently, a treatment-effect estimate should be interpreted together with its confidence interval rather than from sample size alone.
| Quantity | What it contributes |
|---|---|
| Total enrollment | Provides the overall amount of randomized information available. |
| Arm-specific enrollment | Determines the information available for each side of the treatment comparison. |
| Endpoint frequency | Affects the statistical variability of the estimated proportions. |
| Confidence interval | Displays uncertainty around the estimated treatment effect. |
The registry information presented here gives the total enrollment and number of arms but does not specify the arm-specific enrollment allocation or a prespecified power calculation.
14. Limitations
- No posted statistical results: the ClinicalTrials.gov record does not contain a posted statistical analysis for the primary endpoint, so no treatment-effect estimate, confidence interval, p-value, or event count can be reported from the registry record.
- Comparator description: the registry information identifies 2 arms and an endovascular-treatment intervention but does not specify the treatment assigned to the other arm in the information presented here.
- Analysis method not specified: the registry information does not identify the formal statistical test, effect measure, confidence-interval method, or regression model used for the primary endpoint.
- Missing-data rules: the registry information does not specify how participants without an evaluable 90-day mRS assessment are handled.
- Binary endpoint compression: mRS ≤3 simplifies the 0-to-6 scale and therefore does not capture all information contained in the ordinal outcome.
- Unmasked design: the registry describes the study as having no masking. Knowledge of treatment assignment can be relevant to outcomes that involve assessment or clinical management.
- Arm-specific sample sizes: the registry gives total enrollment of 203 and 2 arms but does not provide arm-specific enrollment in the information presented here.
15. Why This Trial Matters Statistically
RESCUE-Japan LIMIT is a useful statistical teaching case because it illustrates how a randomized treatment question can be translated into a precisely defined functional-outcome endpoint. The study combines randomization, a parallel-group design, a clinically ordered outcome scale, and a binary primary threshold measured at a fixed follow-up time.
| Concept | How it appears in RESCUE-Japan LIMIT |
|---|---|
| Randomization | The registry identifies randomized allocation. |
| Parallel-group design | The registry identifies a parallel design with 2 arms. |
| Binary endpoint | The primary outcome is mRS ≤3 at 90 days. |
| Ordered clinical scale | The modified Rankin Scale runs from 0 through 6. |
| Fixed follow-up time | The primary endpoint is assessed at 90 days post-stroke. |
| Absolute treatment effect | A risk difference can express the difference in the proportion achieving mRS ≤3. |
| Relative treatment effect | A risk ratio or odds ratio can express the treatment comparison on a relative scale. |
| Confidence intervals | Precision should accompany any reported treatment-effect estimate. |
| Missing data | A missing 90-day mRS assessment creates an important analysis question. |
| Endpoint dichotomization | The primary endpoint retains a threshold while discarding some of the ordinal information in mRS. |
16. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary statistical question is a comparison of the probability of achieving mRS ≤3 at 90 days between randomized treatment arms. An appropriate analysis would report an effect measure with a confidence interval and a corresponding hypothesis test.
Clinical interpretation
The endpoint is intended to represent functional status after stroke. Its threshold distinguishes participants at or below mRS 3 from those above 3, while the underlying mRS scale contains additional information about the degree of disability.
These two levels of interpretation should remain distinct. Statistical significance addresses evidence against a null hypothesis; clinical interpretation considers what the endpoint represents and how large an observed difference is. A complete report needs both.
17. A Practical Analysis Framework
If a complete 90-day endpoint dataset were available, a transparent analysis would begin with the randomized study population and construct a 2-by-2 table based on whether each participant achieved mRS ≤3.
| mRS ≤3 | mRS >3 | |
|---|---|---|
| Endovascular-treatment arm | Favorable outcome count | Unfavorable outcome count |
| Comparator arm | Favorable outcome count | Unfavorable outcome count |
From this table, the proportions achieving the endpoint could be calculated for each arm. The primary treatment effect could then be expressed on an absolute or relative scale, with a confidence interval and a prespecified hypothesis test.
A more advanced analysis could consider adjustment for prespecified baseline variables or retain the full ordinal mRS outcome. Such analyses should be clearly identified as distinct from the registered binary endpoint rather than presented as though they were the same estimand.
18. What a Complete Results Report Would Contain
For a binary primary endpoint such as mRS ≤3 at 90 days, a statistically complete results presentation would ordinarily contain several complementary pieces of information.
Event counts
The number of participants achieving mRS ≤3 in each randomized arm.
Event proportions
The percentage of participants achieving the primary endpoint in each arm.
Effect estimate
A prespecified measure such as risk difference, risk ratio, or odds ratio.
Confidence interval
An interval quantifying uncertainty around the treatment-effect estimate.
A p-value may also be reported, but it should complement rather than replace the estimated effect and its confidence interval. For a functional-outcome endpoint, the absolute difference between treatment groups can be particularly useful because it communicates the scale of the observed separation directly.
19. Sources
- ClinicalTrials.gov: NCT03702413 — RESCUE-Japan LIMIT.
Continue through the Clinical Biostats knowledge graph
Explore the broader statistical framework used to understand randomized clinical trials, binary endpoints, functional outcomes, treatment effects, and uncertainty.
20. Record Summary
RESCUE-Japan LIMIT is a completed randomized, parallel-group trial with 203 enrolled participants and 2 arms, evaluating endovascular treatment in the setting of acute large vessel occlusion with large ischemic core. Its registered primary endpoint is modified Rankin Scale ≤3 at 90 days, making the central statistical problem a comparison of a binary functional-outcome threshold between randomized groups.
The most important statistical features are the distinction between the underlying 0-to-6 mRS scale and its dichotomized primary endpoint, the choice of an appropriate binary treatment-effect measure, the need to quantify precision with a confidence interval, and the handling of missing 90-day assessments. Because the ClinicalTrials.gov record contains no posted statistical analyses, the registry establishes the design and endpoint but does not provide a numerical treatment-effect result to interpret.