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
SCORPIO-HR was a randomized, double-blind, phase 3, parallel-group trial evaluating S-217622 versus placebo in non-hospitalized participants with SARS-CoV-2 infection. The registered primary endpoint was time to sustained symptom resolution, assessed up to Day 29.
| Feature | SCORPIO-HR |
|---|---|
| Phase | Phase 3 |
| Condition | SARS-CoV-2 Infection |
| Design | Randomized, double-blind, parallel-group |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 2093 |
| Interventions | S-217622 (drug) and placebo (drug) |
| Primary endpoint | Time to Sustained Symptom Resolution |
| Primary endpoint type | Time-to-event |
| Primary endpoint time frame | Up to Day 29 |
| Trial status | Completed |
2. Clinical Question
The central question was whether treatment with S-217622, compared with placebo, changed the time to sustained symptom resolution in non-hospitalized participants with SARS-CoV-2 infection.
Population
Non-hospitalized participants with SARS-CoV-2 infection enrolled in the phase 3 SCORPIO-HR trial.
Intervention
S-217622, also described in the registry results as ensitrelvir.
Comparator
Placebo.
Primary question
Does S-217622 change the time from the start of study intervention to sustained symptom resolution through Day 29?
3. Trial Design
Ensitrelvir
- S-217622
- Randomized treatment assignment
- Included in the modified intent-to-treat analysis after at least one dose and intervention initiation within 3 days of symptom onset
Placebo
- Placebo
- Randomized treatment assignment
- Included in the modified intent-to-treat analysis after at least one dose and intervention initiation within 3 days of symptom onset
The registry identifies the trial as randomized, double-blind, and parallel-group. These design features are important because randomization establishes the basis for a treatment-group comparison, while masking reduces the potential for knowledge of treatment assignment to influence participant or investigator behavior and assessment.
4. Primary Endpoint
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Time to Sustained Symptom Resolution | Time to sustained symptom resolution was defined as the time from start of study intervention to the first day of 2 consecutive days with complete resolution of COVID-19 symptoms on participant self-assessment and alive and without hospitalization for any reason by Day 29. | Up to Day 29 |
Hospitalization was defined as at least 24 hours of acute care in a hospital or similar acute care facility, including emergency rooms, urgent care clinics, or facilities instituted to address medical needs of those with COVID-19.
This definition makes the endpoint more specific than a simple question about whether symptoms improved. A participant had to satisfy the sustained-resolution criterion while also being alive and without hospitalization for any reason by Day 29.
5. Analysis Population
The posted primary analysis used the modified intent-to-treat (mITT) Set: all randomized participants who took at least 1 dose of ensitrelvir or placebo and who started intervention within 3 days of symptom onset.
| Population | Role in the posted analysis |
|---|---|
| Modified intent-to-treat | Primary efficacy analysis population: randomized participants who took at least 1 dose and started intervention within 3 days of symptom onset. |
The distinction between the overall randomized population and the mITT population matters statistically. Randomization is the mechanism that balances prognostic factors between treatment groups in expectation. Restricting the efficacy analysis to participants meeting post-randomization treatment-exposure criteria can alter the population being compared. The mITT definition therefore needs to be kept explicit when interpreting the estimate.
6. Statistical Methodology
Restricted mean symptom duration
The posted analysis compared the difference in restricted mean symptom duration between ensitrelvir and placebo. The analysis notes define this as a comparison of the time, in days, from the start of intervention to sustained symptom resolution.
The reported estimate was −0.6 days. A negative value means that the estimated restricted mean symptom duration was shorter in the ensitrelvir group than in the placebo group.
t-test
The registry reports a 2-sided t-test for the primary endpoint. The statistical method is t-test, with a mean difference as the effect measure.
A t-test is ordinarily associated with comparison of means. Here, the reported analysis expresses the time-to-event endpoint through a restricted mean, allowing the treatment groups to be compared on an average time scale rather than relying on a hazard ratio.
Why restricted mean time is useful for a time-to-event endpoint
Time-to-event data commonly contain participants who do not experience the event during the observation period. Instead of requiring a single hazard ratio to summarize the comparison, restricted mean survival-type measures summarize the expected event-free or event-related time over a specified follow-up horizon.
For this endpoint, the clinically intuitive quantity is time in days to sustained symptom resolution. A mean difference therefore has a direct time interpretation: negative values favor shorter symptom duration in the first treatment group, while positive values favor shorter duration in the comparator group.
Two-sided testing
The posted method is explicitly two-sided. The corresponding P-value was 0.1397. A two-sided test evaluates evidence for a difference in either direction rather than restricting the statistical question to one prespecified direction.
7. Primary Result: Time to Sustained Symptom Resolution
The registry posts one formal statistical analysis for the primary endpoint. The analysis was performed in the modified intent-to-treat set and compared ensitrelvir with placebo using a two-sided t-test of the difference in restricted mean symptom duration.
Difference in restricted mean symptom duration
95% CI: −1.38 to 0.19 · P = 0.1397
Analysis population: modified intent-to-treat set
| Primary endpoint | Ensitrelvir vs placebo |
|---|---|
| Endpoint | Time to Sustained Symptom Resolution |
| Time frame | Up to Day 29 |
| Effect measure | Mean Difference |
| Estimate | −0.6 days |
| 95% confidence interval | −1.38 to 0.19 |
| P-value | 0.1397 |
| Statistical method | t-test, 2 sided |
| Analysis population | Modified intent-to-treat Set |
The point estimate of −0.6 days indicates an estimated shorter restricted mean symptom duration with ensitrelvir relative to placebo. In practical terms, the estimated difference corresponds to less than one day on the restricted mean time scale.
The estimate does not mean that every participant experienced exactly 0.6 fewer days of symptoms. It is a group-level difference in the restricted mean outcome. It also does not establish that every participant had faster symptom resolution.
The 95% confidence interval of −1.38 to 0.19 days includes zero. Thus, the interval includes both a shorter restricted mean symptom duration for ensitrelvir and a small longer duration relative to placebo. The width of the interval reflects uncertainty around the estimated difference.
The P-value of 0.1397 is above the conventional 0.05 threshold. It should not be interpreted as the probability that the treatment works or does not work. Nor does it quantify the size of the observed difference. The estimate and confidence interval provide the more direct description of magnitude and precision.
Reading the confidence interval
The interval from −1.38 to 0.19 days spans zero. A value of zero represents no difference in restricted mean symptom duration between the treatment groups. Values below zero correspond to shorter duration with ensitrelvir, whereas values above zero correspond to shorter duration with placebo under the reported subtraction convention.
8. What the Primary Estimate Means
The sign of the mean difference
The analysis compares ensitrelvir with placebo. The reported estimate is −0.6 days. Because the difference is expressed as ensitrelvir minus placebo, a negative estimate corresponds to a shorter restricted mean symptom duration for ensitrelvir.
Positive mean difference would indicate shorter estimated duration with placebo under the same subtraction convention.
The estimate is not a median
The posted effect measure is a mean difference, not a median difference. Therefore, the −0.6-day estimate should not be described as a 0.6-day difference in median symptom-resolution time.
The estimate is not a hazard ratio
The registry classifies the endpoint as time-to-event, but the posted statistical effect measure is a mean difference derived from restricted mean symptom duration. No hazard ratio is reported in the statistical analysis provided for this endpoint.
The estimate is not an individual-level treatment effect
A mean difference summarizes the contrast between groups. It does not imply that each participant experienced a change of exactly the estimated magnitude. Individual symptom trajectories can vary substantially around a group-level mean.
9. Statistical Methods Explained
Why use a restricted mean for a time-to-event endpoint?
A time-to-event endpoint contains information about when an event occurs and can also contain incomplete observations when the event has not occurred by the end of follow-up. A restricted mean approach summarizes time over a defined observation window and produces an effect measure in the original time unit: days.
What does a mean difference of −0.6 days mean?
Under the reported treatment comparison, the estimated restricted mean symptom duration was 0.6 days shorter with ensitrelvir than with placebo. The negative sign identifies the direction of the difference; the magnitude gives the estimated separation on the day scale.
Why does the confidence interval cross zero?
Zero is the value corresponding to no mean difference. Because the 95% confidence interval extends from −1.38 to 0.19 days, both no difference and modest differences in either direction are compatible with the interval.
Why is the P-value not the effect size?
The P-value and effect estimate answer different questions. The mean difference describes the estimated magnitude and direction of the treatment comparison. The P-value describes the evidence against the statistical null hypothesis under the specified two-sided testing procedure. A smaller P-value does not automatically imply a larger treatment effect.
Why does the mITT definition matter?
The posted analysis was restricted to randomized participants who took at least 1 dose and started intervention within 3 days of symptom onset. This differs conceptually from simply analyzing everyone randomized. When interpreting the result, the population that generated the estimate should therefore be kept explicit.
How is this different from a Cox hazard ratio?
A Cox model summarizes relative event rates through a hazard ratio, whereas a restricted mean analysis summarizes average time over a specified window. The restricted mean difference is expressed directly in days and does not require the treatment effect to be summarized through a proportional-hazards ratio.
What does the two-sided test tell us?
The two-sided test evaluates evidence for a difference in either direction. In this analysis, the reported P-value was 0.1397. The test therefore should be interpreted alongside the estimated −0.6-day difference and its 95% confidence interval rather than in isolation.
10. Time-to-Event Thinking Without a Hazard Ratio
SCORPIO-HR illustrates an important point in survival analysis: a time-to-event endpoint does not require a hazard ratio as its only summary measure.
Event definition
The event is sustained symptom resolution, defined using 2 consecutive days of complete symptom resolution together with being alive and without hospitalization by Day 29.
Time scale
The endpoint measures time from the start of study intervention to sustained symptom resolution.
Restricted mean
The reported analysis converts the time-to-event information into a mean duration over the restricted analysis window.
Interpretability
The resulting treatment contrast is expressed in days, which makes the effect directly interpretable on the symptom-duration scale.
This distinction matters because different time-to-event estimands answer different questions. A hazard ratio describes a relative event-rate contrast under a particular model, while a restricted mean difference describes a difference in average time over a specified horizon. Neither should automatically be substituted for the other.
11. Confidence Interval and Precision
95% confidence interval
Point estimate: −0.6 days
The confidence interval provides the principal measure of statistical precision for the reported mean difference. Its endpoints span a range from a potentially larger shortening of restricted mean symptom duration to a small difference in the opposite direction.
The confidence interval should not be interpreted as saying that there is a 95% probability that the true effect lies between −1.38 and 0.19 days. Rather, under the statistical model and repeated-sampling framework used to construct the interval, the procedure is designed to produce intervals containing the underlying parameter at the stated confidence level over repeated samples.
The interval also gives more information than the P-value alone. It communicates both the direction of the point estimate and the range of effect sizes compatible with the observed data under the analysis framework.
12. Safety: Serious Adverse Events
The registry reports serious adverse events by treatment arm. Eleven participants in the ensitrelvir group were affected among 1037 at risk, compared with 13 participants in the placebo group among 1048 at risk.
| Safety measure | Ensitrelvir | Placebo |
|---|---|---|
| Serious adverse events, affected / at risk | 11 / 1037 | 13 / 1048 |
Why report affected / at risk?
The numerator gives the number of affected participants, while the denominator identifies the number at risk in the reported safety comparison.
Why safety is separate from efficacy
The symptom-resolution endpoint and serious-adverse-event data address different questions. A treatment-effect estimate for symptom duration cannot by itself summarize the safety profile.
The serious-adverse-event counts should be interpreted in the context of the corresponding at-risk populations rather than as isolated event counts.
13. Design Features That Matter Statistically
Randomization
Randomization creates the primary basis for comparing outcomes between S-217622 and placebo. With randomized allocation, differences in observed outcomes can be interpreted within the causal framework of the trial, subject to the analysis population and other design considerations.
Double masking
The trial was double-masked. Masking is particularly relevant for symptom-based outcomes because participants' awareness of treatment assignment could potentially influence self-assessment. The primary endpoint included participant self-assessment of COVID-19 symptoms, making the masking feature statistically relevant to outcome measurement.
Parallel-group design
The trial used a parallel design, so participants were assigned to a treatment group and followed under that assignment rather than sequentially receiving both study interventions as part of a crossover design.
Analysis population
The primary statistical analysis used the modified intent-to-treat set. This population definition should remain attached to the primary estimate because changing the analysis population can change both the estimand and the interpretation of the treatment comparison.
14. Endpoint Construction and Interpretation
The SCORPIO-HR primary endpoint combines several clinical conditions into one time-to-event definition. A participant must have complete resolution of COVID-19 symptoms on self-assessment for 2 consecutive days and must also be alive and without hospitalization for any reason by Day 29.
| Component | Role in the endpoint |
|---|---|
| Start point | Start of study intervention |
| Symptom assessment | Participant self-assessment of COVID-19 symptoms |
| Resolution criterion | Complete resolution for 2 consecutive days |
| Survival requirement | Participant alive by Day 29 |
| Hospitalization requirement | Participant without hospitalization for any reason by Day 29 |
| Time frame | Up to Day 29 |
This construction means that "symptom resolution" should not be treated as a simple binary assessment at one visit. It is a time-based outcome with a sustained-resolution requirement and additional clinical conditions built into the event definition.
15. What the P-value Does — and Does Not — Mean
The reported P = 0.1397 is the result of the specified two-sided t-test for the mean difference in restricted mean symptom duration.
It does not mean there is a 13.97% probability that the treatment effect is zero. It also does not mean there is a 13.97% probability that the observed difference occurred by chance.
Instead, under the null hypothesis and the assumptions of the specified testing framework, it quantifies how compatible the observed result and more extreme results are with that null hypothesis.
The P-value should therefore be read together with the −0.6-day estimate and the −1.38 to 0.19 day 95% confidence interval. The estimate describes the observed direction and magnitude; the interval describes precision; the P-value summarizes evidence against the null under the test.
16. Limitations and Interpretation Issues
- mITT analysis population: the primary result applies to the modified intent-to-treat set defined by randomized treatment assignment, receipt of at least 1 dose, and intervention initiation within 3 days of symptom onset.
- Composite endpoint definition: sustained symptom resolution requires 2 consecutive days of complete symptom resolution and also requires being alive and without hospitalization by Day 29.
- Participant self-assessment: symptom resolution is based on participant self-assessment, making the double-masked design particularly relevant to interpretation.
- Restricted mean estimand: the reported effect is a mean difference in restricted symptom duration, not a hazard ratio or median difference.
- Confidence interval: the 95% interval includes zero, so the estimate is compatible with both a shorter restricted mean symptom duration and a small difference in the opposite direction.
- Time horizon: the registered primary endpoint is assessed up to Day 29. Conclusions about symptom duration beyond this time frame are not part of this endpoint definition.
- Safety interpretation: serious adverse events should be considered separately from the efficacy endpoint and in relation to their corresponding at-risk populations.
17. Why This Trial Matters Statistically
SCORPIO-HR is a useful teaching example because it demonstrates how a clinical trial can use a time-to-event endpoint while reporting its treatment effect on a mean-time scale rather than as a hazard ratio.
| Concept | How it appears in SCORPIO-HR |
|---|---|
| Randomization | Participants were randomized to S-217622 or placebo. |
| Double masking | The trial used double masking, relevant to a participant-reported symptom endpoint. |
| Time-to-event endpoint | Time to sustained symptom resolution was the primary endpoint. |
| Restricted mean | The posted analysis compared restricted mean symptom duration. |
| Mean difference | The treatment effect was reported as a difference in days. |
| Confidence interval | A 95% CI quantified uncertainty around the mean difference. |
| Two-sided testing | The posted t-test was explicitly two-sided. |
| mITT analysis | The primary analysis used the modified intent-to-treat population. |
| Participant-reported outcome | Symptom resolution was based on participant self-assessment. |
| Safety analysis | Serious adverse events were reported as affected participants among those at risk in each arm. |
18. Results in Statistical Context
The primary estimate can be summarized compactly as a treatment-group contrast on the restricted mean symptom-duration scale:
The negative point estimate corresponds to shorter estimated restricted mean symptom duration with ensitrelvir, while the confidence interval includes zero.
This formulation is useful because it keeps the estimand, effect measure, uncertainty, and hypothesis test distinct. The estimand concerns restricted mean symptom duration; the effect measure is a mean difference; the confidence interval quantifies uncertainty around that difference; and the P-value summarizes evidence under the specified two-sided test.
The statistical story is therefore more informative than a simple statement that the P-value was above or below a threshold. The point estimate suggests a shorter restricted mean symptom duration with ensitrelvir, but the confidence interval spans zero and extends from −1.38 to 0.19 days. The result should consequently be understood through the full estimate-and-interval pair rather than through the P-value alone.
19. What the Registry Reports and Does Not Report
The registry reports the primary endpoint definition, the modified intent-to-treat analysis population, the two-sided t-test, the mean-difference effect measure, the estimate of −0.6 days, the 95% confidence interval of −1.38 to 0.19 days, and the P-value of 0.1397.
The registry does not report a median time to sustained symptom resolution in the statistical analysis presented here. It also does not provide a hazard ratio for the primary endpoint in the posted statistical analysis.
That distinction is important because a time-to-event endpoint can be analyzed and summarized in multiple valid ways. The reported SCORPIO-HR analysis uses the restricted mean symptom-duration framework, so the primary result should be interpreted on that scale.
20. Related Tutorials
Learn more about the methods used in this trial:
21. Related Statistical Calculators
22. Sources
- ClinicalTrials.gov: NCT05305547 — SCORPIO-HR.
- Publication: PubMed PMID 39960062.
Continue through Clinical Biostats
Connect the statistical concepts used in this trial to deeper tutorials, statistical calculators, and other clinical-trial analyses.
23. Record Summary
SCORPIO-HR provides a useful example of statistical analysis for a clinical time-to-event endpoint when the primary effect is expressed as a restricted mean difference rather than a hazard ratio. The randomized, double-blind, parallel-group phase 3 design compared S-217622 with placebo in non-hospitalized participants with SARS-CoV-2 infection. The primary endpoint was time to sustained symptom resolution up to Day 29, defined using participant-reported complete symptom resolution over 2 consecutive days together with being alive and without hospitalization by Day 29.
The posted primary analysis used the modified intent-to-treat population and a two-sided t-test, reporting a mean difference of −0.6 days with a 95% confidence interval of −1.38 to 0.19 days and a P-value of 0.1397. The estimate indicates a shorter restricted mean symptom duration with ensitrelvir under the reported subtraction convention, while the confidence interval spans zero. The result is therefore most clearly understood by considering the point estimate, its uncertainty, the analysis population, and the construction of the time-to-sustained-resolution endpoint together.