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
BEST was a randomized, parallel-group clinical trial evaluating endovascular treatment versus standard medical therapy in acute basilar artery occlusion. The registry lists 131 participants and a primary endpoint based on the modified Rankin Scale (mRS) assessed at 90 days from randomization.
| Feature | BEST |
|---|---|
| Trial acronym | BEST |
| NCT ID | NCT02441556 |
| Title | Acute Basilar Artery Occlusion: Endovascular Interventions vs Standard Medical Treatment |
| Condition | Stroke Due to Basilar Artery Occlusion; Acute Cerebrovascular Accidents |
| Phase | NA |
| Status | Completed |
| Enrollment | 131 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
2. Clinical Question
The registry defines the study around acute basilar artery occlusion and compares an endovascular treatment strategy with standard medical therapy. The primary statistical question is whether the randomized groups differ in the proportion of participants achieving a favorable functional outcome on the modified Rankin Scale at 90 days from randomization.
Population
Participants with stroke due to basilar artery occlusion or acute cerebrovascular accidents represented by the registry's condition definitions.
Intervention
Endovascular treatment, listed in the registry as a device intervention.
Comparator
Standard medical therapy, listed in the registry as an other intervention.
Primary question
Does endovascular treatment differ from standard medical therapy with respect to favorable functional outcome at 90 days from randomization?
3. Trial Design
Endovascular treatment
- Intervention type: device
- Randomized treatment strategy
- Primary endpoint assessed at 90 days from randomization
Standard medical therapy
- Intervention type: other
- Randomized comparator strategy
- Primary endpoint assessed at 90 days from randomization
4. Trial Timeline
Study start
The registry lists January 2015 as the study start date.
Enrollment
The registry lists an enrollment of 131 participants in the randomized trial.
Primary completion
The registry lists September 27, 2017 as the primary completion date.
Study status
The registry identifies the study status as completed.
5. Primary Endpoint
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Modified Rankin Scale (mRS) | A score of 0-3 will be considered as the favorable outcome. | At 90 days from randomization |
The mRS is a functional outcome scale used after stroke. Rather than treating the endpoint as a continuous measurement, the registry explicitly defines a favorable outcome as an mRS score of 0-3. Statistically, that converts the original ordered clinical scale into a binary endpoint: favorable outcome versus an mRS score outside the favorable range.
The analysis therefore focuses on whether the probability of a favorable functional outcome differs between the randomized treatment groups at the specified 90-day assessment.
6. Planned Analysis
The registry identifies the modified Rankin Scale as the primary endpoint, with favorable outcome defined as a score of 0-3 at 90 days from randomization. Results have not been posted on ClinicalTrials.gov.
For a binary endpoint of this type, a conventional primary comparison would generally summarize the proportion of participants achieving a favorable outcome in each randomized group and compare those proportions using an appropriate two-group method. Depending on the prespecified statistical analysis plan and the observed cell counts, common approaches include a Pearson chi-square test or Fisher's exact test, together with an effect measure such as a risk difference, risk ratio, or odds ratio and a confidence interval.
The most clinically transparent presentation would usually begin with the two favorable-outcome proportions. A risk difference answers the absolute question—how many additional favorable outcomes are observed per 100 participants under one treatment strategy relative to the other—whereas a risk ratio describes the favorable-outcome probability on a relative scale. An odds ratio uses odds rather than probabilities and therefore should not be read as though it were a risk ratio when favorable outcomes are common.
What would be estimated?
The primary treatment contrast would be the difference in favorable functional outcome between the two randomized groups at 90 days.
What would be tested?
The formal hypothesis would concern whether the randomized treatment groups have different favorable-outcome probabilities at the specified assessment time.
Why use an effect estimate?
A p-value addresses evidence against a null hypothesis, while a risk difference, risk ratio, or odds ratio describes the magnitude and direction of the treatment contrast.
Why include a confidence interval?
A confidence interval communicates statistical precision around the estimated treatment contrast and helps distinguish a precise estimate from one with substantial uncertainty.
7. Statistical Methodology
Randomization and causal comparison
Randomization is the central design feature that permits a comparison of outcomes between treatment strategies while reducing systematic differences in baseline characteristics in expectation. The important statistical unit is the randomized participant, not merely the treatment actually received after randomization.
In a randomized parallel-group trial, the primary analysis would ordinarily preserve the randomized comparison. This is commonly expressed through the intention-to-treat principle, under which participants are analyzed according to their assigned treatment group. The registry information available here does not specify an analysis population or state whether an intention-to-treat analysis was used.
Binary interpretation of the mRS endpoint
The registry does not designate the mRS as a fully ordinal primary analysis. Instead, it specifies a threshold: scores from 0-3 are considered favorable. That distinction is important because dichotomization discards information about the exact position within the scale. A participant with an mRS of 0 and a participant with an mRS of 3 are both counted as favorable even though their functional states differ.
A positive risk difference would indicate a higher observed favorable-outcome probability in the endovascular group; a negative value would indicate a higher observed favorable-outcome probability in the standard-medical-therapy group.
Risk ratio
A risk ratio compares the probability of favorable outcome between the treatment groups. Unlike an odds ratio, it remains directly expressed as a ratio of probabilities, making it intuitive when the outcome itself is the clinical event of interest.
A risk ratio of 1 corresponds to equal favorable-outcome probabilities. Values above 1 indicate a higher favorable-outcome probability in the numerator group, while values below 1 indicate a lower probability.
Odds ratio
An odds ratio compares the odds of favorable outcome rather than the probabilities themselves. It can be useful in binary-outcome analysis, particularly in logistic regression, but it is not interchangeable with the risk ratio.
Here, pE represents the favorable-outcome probability in the endovascular group and pM the corresponding probability in the standard-medical-therapy group.
Confidence intervals
A confidence interval around a treatment effect describes statistical uncertainty under the selected model and sampling framework. It should not be interpreted as a range containing a fixed percentage of individual patients' treatment effects. The width of the interval is particularly important in a trial with a finite enrollment because fewer observations generally provide less precision than a substantially larger randomized population.
P-values
If a formal hypothesis test were used, the p-value would quantify how unusual the observed data would be under the specified null hypothesis and statistical model. It would not measure the size of the treatment effect. A small p-value can accompany a modest effect in a sufficiently large study, while a potentially meaningful effect can fail to produce a small p-value when statistical precision is limited.
8. Statistical Methods Explained
Why was the mRS converted into a favorable-outcome endpoint?
The registry explicitly defines an mRS score of 0-3 as favorable. This creates a clinically interpretable binary endpoint, allowing the trial to ask a direct question about the proportion of participants achieving the prespecified functional-outcome threshold.
What information is lost by dichotomizing the mRS?
Dichotomization simplifies interpretation but removes distinctions within the two resulting categories. All scores from 0 through 3 receive the same favorable-outcome classification for the primary endpoint, even though the underlying ordinal scale contains more information.
Why does randomization matter for the statistical analysis?
Randomization creates the basis for comparing outcomes between treatment strategies without relying solely on observational adjustment. If the randomized comparison is preserved in the primary analysis, differences in outcomes can be interpreted in relation to treatment assignment rather than simply to differences between people who happened to receive different therapies.
What would a risk difference tell us?
A risk difference expresses the absolute separation between the two treatment-group probabilities of favorable outcome. It is especially useful for clinical interpretation because it directly answers how much the probability of the favorable endpoint differs between groups.
How is a risk ratio different from an odds ratio?
A risk ratio compares probabilities directly. An odds ratio compares odds, where odds are the probability of the event divided by the probability of the event not occurring. When outcomes are common, the numerical value of an odds ratio can differ substantially from the corresponding risk ratio, so the two measures should not be described interchangeably.
Why is the 90-day time frame important?
The registry defines the primary endpoint at 90 days from randomization. The timing is part of the endpoint definition: changing the assessment time would create a different statistical estimand because functional status can change over follow-up.
Would an ordinal analysis be possible?
Yes. The mRS itself is ordinal, so an analysis can in principle use the full ordering of the categories rather than collapsing them into favorable versus not favorable. However, the registry definition posted on ClinicalTrials.gov for the primary endpoint specifically establishes 0-3 as the favorable category. A full ordinal analysis should therefore be distinguished from the registered binary endpoint rather than substituted for it.
9. Interpretation of the Primary Endpoint
The primary endpoint is a functional outcome measured using the modified Rankin Scale at 90 days from randomization. The registry defines scores of 0-3 as favorable, so the central comparison is the proportion of randomized participants reaching that threshold in the endovascular-treatment group versus the standard-medical-therapy group.
The endpoint does not mean that every participant with a favorable score had the same degree of functional recovery. An mRS of 0 and an mRS of 3 are both classified as favorable under the registered definition, despite representing different points on the underlying scale.
A treatment effect should therefore be reported with both its direction and its uncertainty. A confidence interval around a risk difference, risk ratio, or odds ratio would describe the statistical precision of the estimated group contrast; it would not describe the distribution of outcomes for individual patients.
A p-value, if reported from a formal test, would address the null hypothesis specified for that analysis. It would not quantify the clinical importance of the treatment effect and should be interpreted alongside the effect estimate and confidence interval.
10. Analysis Population and Missing Data
The registry information identifies enrollment and randomization but does not specify the primary efficacy analysis population or the handling of missing 90-day mRS assessments.
For a randomized binary endpoint, the analysis population is consequential. An intention-to-treat analysis generally retains the benefits of randomization by analyzing participants according to their randomized assignment. A per-protocol or evaluable-population analysis can answer a somewhat different question because participants with protocol deviations, treatment discontinuation, or missing assessments may be handled differently.
Missing 90-day mRS observations also require an explicit strategy. A complete-case analysis can reduce the effective sample and may introduce bias if missingness is related to prognosis or treatment. Other approaches can use assumptions about the missing outcomes. The appropriate method depends on the prespecified analysis plan and the mechanism and amount of missingness; the registry information summarized here does not specify such a method.
11. What the Trial Design Does — and Does Not — Establish
What randomization establishes
Randomization provides the design framework for a comparative treatment analysis and reduces systematic allocation differences in expectation.
What unmasked means
The registry identifies the trial as unmasked. Treatment assignment was therefore not described as masked in the registry design information.
What the parallel design means
The two treatment strategies form concurrent groups rather than a crossover sequence in which each participant receives both interventions.
What the endpoint means
The primary outcome is functional status at a defined follow-up time, summarized using a prespecified favorable mRS threshold.
12. Limitations
- No posted statistical analyses: the ClinicalTrials.gov record does not contain posted statistical analyses for the primary endpoint.
- Endpoint dichotomization: defining mRS 0-3 as favorable simplifies interpretation but discards some information contained in the underlying ordinal scale.
- Analysis specification: the registry information does not identify the exact hypothesis test, effect measure, confidence-interval method, or model used for the primary endpoint.
- Analysis population: the registry information does not specify whether the primary analysis used intention-to-treat, modified intention-to-treat, per-protocol, or another population definition.
- Missing data: the registry information does not specify how missing 90-day mRS assessments were handled.
- Unmasked design: the absence of masking can be relevant when outcomes involve assessment of functional status, because knowledge of treatment assignment can potentially affect aspects of outcome assessment or participant management.
- Enrollment: the randomized trial enrolled 131 participants, so statistical precision is inherently dependent on the number of participants contributing valid primary-endpoint information and the distribution of outcomes between groups.
- External validity: the registry condition definitions describe the target clinical setting, but the information summarized here does not provide a detailed baseline population profile from which broader generalizability can be assessed.
13. Why This Trial Matters Statistically
BEST is a useful teaching example because a clinically meaningful stroke outcome has been translated into a clearly defined binary statistical endpoint within a randomized parallel-group design. The trial illustrates how treatment assignment, endpoint definition, follow-up timing, and statistical effect measures interact.
| Statistical concept | How it appears in BEST |
|---|---|
| Randomization | Participants were allocated randomly to the two treatment strategies. |
| Parallel-group design | Endovascular treatment and standard medical therapy were evaluated as separate concurrent arms. |
| Binary endpoint | The mRS was converted into favorable versus not favorable using the registered 0-3 threshold. |
| Fixed follow-up | The primary endpoint is assessed at 90 days from randomization. |
| Risk difference | Provides an absolute comparison of favorable-outcome probabilities between groups. |
| Risk ratio | Provides a relative comparison of favorable-outcome probabilities. |
| Odds ratio | Compares the odds of favorable outcome rather than the probabilities directly. |
| Confidence interval | Would quantify uncertainty around the selected treatment-effect estimate. |
| Hypothesis testing | A formal test could assess the null hypothesis of no difference in favorable-outcome probability. |
| Missing data | Handling of absent 90-day mRS assessments can affect the analysis population and interpretation. |
14. A Statistical Walk-Through of the Primary Endpoint
The analysis can be understood as a sequence of increasingly informative questions.
Step 1 — Define the outcome. The endpoint is the modified Rankin Scale measured at 90 days from randomization.
Step 2 — Apply the threshold. Scores of 0-3 are classified as favorable. This transforms the endpoint into a binary variable suitable for a two-group proportion comparison.
Step 3 — Compare randomized groups. The favorable-outcome probability is evaluated separately for the endovascular-treatment group and the standard-medical-therapy group.
Step 4 — Quantify the difference. An absolute risk difference and/or relative effect measure can communicate how far apart the two treatment groups are.
Step 5 — Quantify uncertainty. A confidence interval indicates how precisely the treatment effect has been estimated. A hypothesis test can then address the prespecified null hypothesis, but the p-value should not replace the effect estimate.
15. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary endpoint is a binary functional-outcome measure derived from the mRS at a prespecified time. The appropriate treatment contrast concerns the probability of favorable outcome between the randomized groups, accompanied by an uncertainty measure.
Clinical interpretation
The endpoint is intended to capture functional status after acute basilar artery occlusion. Its meaning depends on the patient's position on the mRS, although the registered primary analysis groups scores 0-3 together as favorable.
16. Understanding the mRS Threshold
The modified Rankin Scale is ordered, with different levels representing different degrees of functional outcome. The registry's definition establishes a boundary at 0-3 for the primary favorable-outcome classification.
| Endpoint component | Statistical meaning |
|---|---|
| mRS 0-3 | Counted as a favorable outcome for the registered primary endpoint. |
| mRS outside 0-3 | Not counted as a favorable outcome for the registered primary endpoint. |
| 90 days | The specified time point for primary endpoint assessment. |
| Randomization | The time origin from which the 90-day endpoint is defined. |
This threshold creates a clinically understandable endpoint but also creates a statistical tradeoff. A binary endpoint is easier to summarize than a full ordinal distribution, yet two patients assigned the same binary status can have meaningfully different mRS scores. That tradeoff should be remembered whenever the primary analysis is interpreted.
17. Trial Status and Registry Record
| Registry field | Recorded value |
|---|---|
| NCT ID | NCT02441556 |
| Acronym | BEST |
| Status | Completed |
| Start date | 2015-01 |
| Primary completion date | 2017-09-27 |
| Lead sponsor | Xinfeng Liu |
| Sponsor type | OTHER |
| Enrollment | 131 |
| Number of arms | 2 |
18. Sources
- ClinicalTrials.gov: BEST — NCT02441556.
Continue through the Clinical Biostats knowledge graph
Clinical trial endpoints become easier to interpret when their definitions, estimands, effect measures, confidence intervals, and hypothesis tests are considered together.
19. Record Summary
BEST is a randomized, parallel-group trial with 131 enrolled participants comparing endovascular treatment with standard medical therapy for acute basilar artery occlusion. Its registered primary endpoint is the modified Rankin Scale at 90 days from randomization, with a score of 0-3 defined as a favorable outcome. Statistically, this creates a binary functional endpoint whose natural treatment contrast is the difference in favorable-outcome probability between the randomized groups. The most informative analysis would pair that treatment effect with a confidence interval and an appropriately specified hypothesis test, while clearly distinguishing the registered endpoint from the underlying ordinal mRS scale.