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
SOUL was a randomized, parallel-group, quadruple-masked phase 3 treatment trial in patients with type 2 diabetes. The registry identifies oral semaglutide and placebo as the two interventions and reports a primary cardiovascular endpoint based on the first occurrence of cardiovascular death, non-fatal myocardial infarction, or non-fatal stroke.
| Feature | SOUL |
|---|---|
| Trial name | SOUL |
| ClinicalTrials.gov identifier | NCT03914326 |
| Phase | Phase 3 |
| Condition | Diabetes Mellitus, Type 2 |
| Therapeutic area | Cardiovascular; Endocrinology & Metabolism |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 9651.0 |
| Lead sponsor | Novo Nordisk A/S |
| Sponsor type | Industry |
| Status | Completed |
| Start | 2019-06-17 |
| Primary completion | 2024-08-23 |
2. Clinical Question
The central question was whether oral semaglutide, compared with placebo, affected the time from randomization to the first occurrence of a major adverse cardiovascular event in patients with type 2 diabetes.
Population
Participants with Diabetes Mellitus, Type 2 enrolled in the phase 3 SOUL trial.
Intervention
Oral semaglutide.
Comparator
Placebo.
Primary question
How does oral semaglutide compare with placebo for time to first EAC-confirmed MACE during the in-trial observation period?
3. Trial Design
Oral Semaglutide
- Oral semaglutide
- Randomized treatment assignment
- Follow-up for cardiovascular outcomes
Placebo
- Placebo
- Randomized treatment assignment
- Follow-up for cardiovascular outcomes
The registry describes the study as quadruple-masked, meaning masking was incorporated broadly into the trial conduct. Randomization provides the principal design mechanism for comparing the treatment groups, while masking can reduce the influence of knowledge of treatment assignment on trial conduct and outcome assessment.
4. Primary Endpoint
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Major adverse cardiovascular event (MACE) | Number of participants from randomization to first occurrence of a major adverse cardiovascular event (MACE), a composite endpoint consisting of cardiovascular (CV) death, non-fatal myocardial infarction, or non-fatal stroke. The event was required to be confirmed by the event adjudication committee. | From randomisation (week 0) up to week 265 |
The registry defines the endpoint around the first occurrence of the composite event. The three component outcomes are combined into one primary endpoint: cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke.
5. Statistical Methodology
Time-to-event analysis
Although the registered primary endpoint is classified as binary in the registry, the posted formal analysis treats the outcome as a time-to-event endpoint. The analysis compares the time from randomization to first EAC-confirmed MACE rather than simply classifying participants as having or not having an event without regard to follow-up time.
Cox proportional-hazards model
The registry reports a Cox regression analysis with treatment as a categorical fixed factor. The comparison is oral semaglutide versus placebo, and the effect measure is the hazard ratio.
The formal analysis population was the full analysis set (FAS), defined as all unique randomized participants grouped according to the treatment assigned at randomization.
Analysis population
The FAS included all unique randomized participants and analyzed them according to their randomized treatment assignment. This is important because the primary treatment comparison remains anchored to randomization rather than being redefined according to treatment actually received after randomization.
Hypothesis type
The registry identifies the hypothesis as superiority. Thus, the statistical question is whether the randomized treatment comparison provides evidence of a difference in the time-to-event outcome, rather than whether semaglutide meets a prespecified non-inferiority margin.
Effect measure
The effect measure is the hazard ratio (HR). For this analysis, an HR below 1 corresponds to a lower estimated instantaneous event rate in the oral-semaglutide group relative to placebo under the Cox model.
6. Primary Result: Major Adverse Cardiovascular Events
Hazard ratio for first MACE
95% CI: 0.77–0.96 · P = 0.0028
Two-sided confidence interval · Superiority analysis
| Analysis feature | Reported result |
|---|---|
| Endpoint | Time from randomization to first EAC-confirmed MACE |
| Comparison | Oral semaglutide vs placebo |
| Analysis population | Full analysis set |
| Model | Cox proportional-hazards model |
| Effect measure | Hazard ratio |
| Hazard ratio | 0.86 |
| 95% confidence interval | 0.77–0.96 |
| P-value | 0.0028 |
| Hypothesis | Superiority |
An HR of 0.86 means that, under the fitted Cox model, the estimated instantaneous rate of experiencing a first MACE was approximately 14% lower in the oral-semaglutide group than in the placebo group over the analyzed follow-up.
The HR does not mean that exactly 14% fewer participants experienced MACE, that 14% of participants benefited, or that every individual participant had a 14% reduction in cardiovascular risk. The hazard ratio is a relative model-based measure of event rates over time.
The 95% CI of 0.77–0.96 describes uncertainty around the estimated hazard ratio. Because the interval is entirely below 1, the interval is consistent with a lower estimated hazard in the oral-semaglutide group under the model.
The p-value of 0.0028 addresses the statistical evidence against the null hypothesis under the specified testing framework. It does not measure the size or clinical importance of the treatment effect. Effect size is described by the hazard ratio and its confidence interval.
The interpretation also depends on the Cox proportional-hazards framework and its assumptions. A single hazard ratio is most naturally interpreted when the relative hazards are reasonably represented by a common ratio over the analyzed time period.
7. Understanding the MACE Endpoint Statistically
A composite endpoint combines several clinically distinct events into a single analysis. In SOUL, the composite consists of cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke.
Why use a composite?
A composite can capture the first occurrence of several related clinical outcomes within one prespecified endpoint. Statistically, this allows time to the first qualifying event to be analyzed as a single time-to-event outcome.
Why first occurrence?
Once a participant experiences the first qualifying MACE, that participant has reached the primary endpoint. The endpoint therefore focuses on time to first event rather than repeated event counts.
Why adjudication matters
The registry specifies EAC confirmation. An event adjudication committee provides a structured process for determining whether reported clinical events satisfy the prespecified endpoint definitions.
Why time matters
Participants may be followed for different lengths of time. Time-to-event methods use information about both whether an event occurred and when it occurred, while accommodating incomplete observation through censoring.
8. Hazard Ratio: What the Estimate Tells Us
For SOUL, the reported HR is 0.86. Values below 1 indicate a lower estimated event rate in the numerator group under the Cox model.
The most important distinction is between a hazard ratio and an absolute probability. A hazard ratio compares event rates over time. It does not directly state the probability that an individual participant will experience MACE by a particular week.
The confidence interval adds an essential second dimension. An estimate of 0.86 is only one point within a range of values compatible with the statistical uncertainty represented by the 95% confidence interval. The interval of 0.77–0.96 is therefore more informative than the point estimate alone.
The result can also be expressed directionally: an HR of 0.86 corresponds to an estimated 14% lower hazard because 1 − 0.86 = 0.14. That derived percentage is a transformation of the reported hazard ratio, not a separate event-rate estimate.
9. Why the P-value Is Not the Effect Size
The reported p-value is 0.0028. A p-value quantifies how compatible the observed data are with a specified null hypothesis under the statistical model and testing framework. It is not a measure of how large the treatment effect is.
| Quantity | What it communicates | What it does not communicate |
|---|---|---|
| HR 0.86 | Relative treatment effect on the hazard scale | The absolute probability of MACE for an individual participant |
| 95% CI 0.77–0.96 | Uncertainty around the estimated HR | The range of effects experienced by individual patients |
| P = 0.0028 | Evidence against the null under the specified testing framework | The magnitude or clinical importance of the effect |
10. Full Analysis Set and Randomization
The registry defines the FAS as including all unique randomized participants, grouped according to the treatment assigned at randomization. This is closely connected to the purpose of randomization: assignment creates the basis for a comparison in which treatment groups can be interpreted through the randomized allocation rather than through post-randomization treatment selection.
Analyzing participants according to their randomized assignment also avoids changing the treatment comparison simply because treatment exposure or subsequent follow-up differs between individuals. For a primary efficacy question, that preserves the connection between the statistical analysis and the randomized trial design.
11. Censoring and In-Trial Observation
Time-to-event analyses require a rule for determining how long each participant contributes observable follow-up. The registry defines the in-trial observation period from randomization until the first applicable endpoint among follow-up, withdrawal of consent, last contact for a participant lost to follow-up, or death.
This matters because a participant who has not experienced MACE by the end of observable follow-up does not necessarily provide evidence of being permanently event-free. Instead, the participant contributes information up to the point at which observation ends.
The Cox model uses this longitudinal event-time structure rather than reducing the analysis to a simple count of participants with and without MACE.
12. Statistical Methods Explained
Why was a Cox proportional-hazards model used?
The primary endpoint is fundamentally about the time from randomization to first MACE. Cox regression is designed for this type of time-to-event outcome and produces a hazard ratio that compares the instantaneous event rates between treatment groups.
What does an HR of 0.86 mean?
It means that the estimated instantaneous rate of first MACE was approximately 14% lower with oral semaglutide than with placebo under the fitted Cox model. It does not mean that 14% of participants avoided MACE or that the absolute event probability was reduced by 14 percentage points.
Why is the confidence interval important?
The point estimate is only one estimate of the treatment effect. The 95% CI of 0.77–0.96 shows the statistical uncertainty surrounding the HR under the model and sampling framework. A narrower interval would indicate greater precision than a wider interval, all else equal.
Why does the p-value not tell us how large the effect is?
The p-value measures evidence against a null hypothesis within the specified statistical framework. A very small p-value can accompany a modest effect in a sufficiently large study, while a clinically meaningful effect can produce a less impressive p-value in a small study. The HR and confidence interval are therefore needed to describe the magnitude and precision of the observed treatment effect.
Why analyze participants according to randomized treatment assignment?
The FAS groups unique randomized participants according to treatment assigned at randomization. This keeps the primary comparison connected to the randomized design and avoids redefining treatment groups according to later treatment exposure.
What is the proportional-hazards assumption?
The Cox model expresses treatment effects through a hazard ratio. The standard proportional-hazards interpretation assumes that the relative hazard is adequately represented by a common ratio over time. If that assumption is substantially violated, a single HR can become less descriptive of how treatment effects vary across follow-up.
13. Safety Results
The registry reports serious adverse events by treatment arm as affected participants divided by participants at risk.
| Safety measure | Oral Semaglutide | Placebo |
|---|---|---|
| Serious adverse events | 2312/4825 | 2427/4825 |
The reported figures correspond to affected participants over participants at risk in each arm. They should be interpreted as a safety summary separate from the primary MACE efficacy analysis.
The serious-adverse-event figures describe the number of affected participants relative to the number at risk in each treatment group. They are not the same statistical quantity as the primary hazard ratio, and they should not be combined into a single overall efficacy-safety score.
The primary MACE result addresses time to a prespecified cardiovascular composite, whereas serious adverse events represent a broader safety category. These endpoints answer different questions and require separate interpretation.
14. What the Registry Reports — and What It Does Not
The registry reports 25 outcome measures and 1 statistical analysis, with one primary-endpoint analysis including an estimate and confidence interval. The formal statistical analysis posted for the primary endpoint is the Cox proportional-hazards model described above.
| Element | Registry status |
|---|---|
| Results posted | Yes |
| Outcome measures posted | 25 |
| Statistical analyses posted | 1 |
| Primary-endpoint analyses | 1 |
| Primary analyses with estimate + CI | 1 |
| Formal method | Cox proportional-hazards model |
| Effect measure | Hazard ratio |
| Hypothesis type | Superiority |
The registry does not report median survival, subgroup estimates, a baseline characteristics table, a forest plot, or additional formal statistical analyses in the information summarized here. Those elements are therefore not necessary to interpret the posted primary analysis.
15. Important Limitations and Interpretation Issues
- Composite endpoint: MACE combines cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke. A composite treatment effect does not by itself establish that each component has the same magnitude of effect.
- First-event analysis: the registered endpoint concerns the first occurrence of a qualifying MACE, so the primary analysis is not a repeated-event analysis.
- Hazard-ratio interpretation: the HR is model-based and depends on the proportional-hazards framework. A single HR can be less descriptive if relative hazards vary materially over time.
- Censoring: participants may have incomplete observable follow-up because of withdrawal, loss to follow-up, or the registry-defined end of observation. Time-to-event methods account for such incomplete observation through their censoring framework.
- Absolute risk: the posted primary analysis provides a relative hazard measure, but the summarized registry data do not provide the event counts needed here to derive absolute MACE risks at specific time points.
- Endpoint classification: the registry classifies the registered primary endpoint as binary while the posted formal analysis is a time-to-event Cox analysis. The statistical method therefore incorporates event timing rather than treating the endpoint as a simple fixed-time binary proportion.
- Safety versus efficacy: serious adverse events and MACE are distinct outcomes and should be interpreted separately rather than collapsed into one summary measure.
- Generalizability: the trial's applicability to other populations depends on the eligibility criteria and characteristics of the enrolled study population, which are not detailed in the summarized registry information.
16. Why This Trial Matters Statistically
SOUL is a useful teaching case because the registry illustrates how a large randomized cardiovascular-outcomes trial can be reduced to a clearly defined time-to-event question and analyzed with a Cox proportional-hazards model.
| Concept | How it appears in SOUL |
|---|---|
| Randomization | Participants were randomized in a parallel-group phase 3 treatment trial. |
| Masking | The study was quadruple-masked. |
| Time-to-event endpoint | Time from randomization to first EAC-confirmed MACE. |
| Composite endpoint | CV death, non-fatal myocardial infarction, and non-fatal stroke. |
| Cox model | Used to compare time to first MACE between oral semaglutide and placebo. |
| Hazard ratio | The reported treatment effect was HR 0.86. |
| Confidence interval | The 95% CI was 0.77–0.96. |
| Superiority testing | The registry identifies the hypothesis type as superiority. |
| Full analysis set | Unique randomized participants were grouped according to treatment assigned at randomization. |
| Safety analysis | Serious adverse events were reported separately by treatment arm. |
17. Interpreting the Complete Statistical Story
The primary analysis compares oral semaglutide with placebo for time to first EAC-confirmed MACE using a Cox proportional-hazards model. The reported HR of 0.86 indicates a lower estimated instantaneous event rate in the oral-semaglutide group, with a two-sided 95% CI of 0.77–0.96 and a p-value of 0.0028.
The HR does not provide an absolute event probability, does not state the number of participants who benefited, and does not imply that every component of the composite endpoint experienced the same treatment effect. Those are separate statistical questions.
Because the FAS retains randomized treatment assignment, the primary comparison remains linked to the allocation generated by the randomized trial design. This is an important part of interpreting the treatment effect as a randomized comparison rather than as an observational comparison of people who happened to receive different treatments.
MACE is not analyzed merely as a yes-or-no characteristic. The registry's formal analysis uses the time from randomization to the first qualifying event. That makes follow-up time and censoring integral parts of the statistical analysis.
18. Related Tutorials
Learn more about the methods used in this trial:
19. Related Calculators
20. Sources
- ClinicalTrials.gov: SOUL, NCT03914326.
- PubMed: PMID 41879791.
- PubMed: PMID 41627802.
- PubMed: PMID 41380027.
- PubMed: PMID 41099689.
- PubMed: PMID 40162642.
Continue through the Clinical Biostats knowledge graph
Connect the trial's endpoint and statistical methods to deeper tutorials, calculators, and related clinical-trial analyses.
21. Record Summary
SOUL provides a clear example of a randomized cardiovascular time-to-event analysis. The trial enrolled 9651.0 participants in a phase 3, parallel, quadruple-masked design comparing oral semaglutide with placebo. Its primary endpoint was time from randomization to first EAC-confirmed MACE, defined as cardiovascular death, non-fatal myocardial infarction, or non-fatal stroke, with follow-up specified from randomisation at week 0 up to week 265.
The formal analysis used the full analysis set and a Cox proportional-hazards model with treatment as a categorical fixed factor. The reported hazard ratio was 0.86, with a two-sided 95% confidence interval of 0.77–0.96 and a p-value of 0.0028. Statistically, the result describes a lower estimated hazard of first MACE under the fitted model; it should not be translated directly into an absolute risk reduction or into a claim that the same effect applies to every component of the composite endpoint.
The registry also reports serious adverse events separately, with 2312/4825 affected participants in the oral-semaglutide group and 2427/4825 in the placebo group. This safety information addresses a different outcome from the primary cardiovascular efficacy analysis and should therefore be interpreted independently.