The SOUL trial already made its headline. In 2025, researchers reported that once-daily oral semaglutide, the GLP-1 medicine sold for type 2 diabetes as Rybelsus, reduced the risk of a first major cardiovascular event by 14% compared with placebo in people with type 2 diabetes and established cardiovascular or kidney disease. But heart disease rarely stops after one event. A new secondary analysis published November 11 in JAMA Cardiology counted every major cardiovascular event during the trial, not just the first, and found fewer of them in the semaglutide group. The difference was statistically significant in the combined analysis. Whether the medicine specifically prevents second or third events remains less certain.
That distinction is important. The researchers recorded 108 fewer total major cardiovascular events among participants assigned to oral semaglutide than among those assigned to placebo. Their statistical model estimated a 14% reduction in cumulative event burden. But the trial was designed primarily to test time to a first event, and the separate results for recurrent events were inconclusive. The new analysis strengthens the evidence for overall cardiovascular benefit without proving every part of the story.
What the SOUL Trial Tested
SOUL was a randomized, double-blind, placebo-controlled human trial conducted at 444 centers in 33 countries between June 2019 and September 2024. It enrolled 9,650 adults with type 2 diabetes and atherosclerotic cardiovascular disease, chronic kidney disease, or both. The median participant age was 66, and approximately 71% were men. Participants received once-daily oral semaglutide, with doses escalated to a maximum of 14 milligrams, or a placebo, in addition to standard medical care.
Participants were followed for a median of 49.5 months, just over four years. The original study, published in the New England Journal of Medicine in 2025, measured time to the first three-point major adverse cardiovascular event, known as 3-P MACE. That composite includes nonfatal myocardial infarction, or heart attack; nonfatal stroke; and cardiovascular death.
In the original analysis, 579 of the 4,825 people assigned to semaglutide experienced a first major cardiovascular event, compared with 668 of the 4,825 participants receiving placebo. Those figures correspond to 12.0% and 13.8%, respectively, with a hazard ratio of 0.86. The result established a statistically significant reduction in the risk of a first event in this high-risk population.
But a standard time-to-first-event analysis does not capture the full history of someone who has a nonfatal heart attack, survives, and later has a stroke. For the primary outcome, the counting stops at the first qualifying event. A recurrent-event analysis follows the additional events as well.
A first heart attack can end the count in a clinical trial analysis. It rarely ends a patient’s cardiovascular risk.
Counting First and Repeat Events
The new analysis, led by Nikolaus Marx of RWTH Aachen University and colleagues from academic institutions and Novo Nordisk, revisited SOUL’s cardiovascular outcomes using a statistical method that accounts for recurrent events and for death from causes outside the cardiovascular composite.
Across the trial, investigators recorded 1,474 major cardiovascular events. Of those, 1,247 were first events and 227 were subsequent events, meaning additional events experienced by participants who had already had at least one qualifying outcome.
The semaglutide group experienced 683 total events, comprising 579 first events and 104 subsequent events. The placebo group experienced 791 total events, comprising 668 first events and 123 subsequent events. The difference was 108 total events, including 19 fewer subsequent events in the semaglutide group.
Using a Ghosh-Lin recurrent-event model, the researchers estimated a mean ratio of 0.86, with a 95% confidence interval of 0.76 to 0.97 and a P value of .01. That represents a 14% lower modeled cumulative burden of major cardiovascular events with oral semaglutide compared with placebo.
The distinction between total and repeat events becomes clearer when the researchers separate them. The hazard ratio for a second event was 0.85, but its 95% confidence interval ranged from 0.63 to 1.13. For a third event, the estimate was 0.72, with a still wider confidence interval of 0.38 to 1.37. Neither result was statistically significant on its own.
Only 180 participants, approximately 1.9% of the study population, experienced more than one major cardiovascular event. Of those, 83 were in the semaglutide group and 97 in the placebo group. The relatively small number of people experiencing multiple events limits how confidently researchers can estimate treatment effects specifically after a first event.

The investigators also conducted an exploratory analysis that added non-cardiovascular deaths to the composite. That produced 910 total events in the semaglutide group and 1,048 in the placebo group, a difference of 138, with an estimated mean ratio of 0.87. This was a separate, post hoc analysis and should not be treated as the trial’s original primary endpoint.
What the Cardiovascular Benefit Means for Rybelsus
The findings offer another reason to take oral semaglutide seriously as a cardiovascular medicine, not only as a drug for lowering blood glucose. In SOUL, a pill produced a significant reduction in the original first-event outcome and a similar estimated reduction when the researchers counted all qualifying cardiovascular events.
That matters for people living with type 2 diabetes and existing cardiovascular disease. Their risks do not reset after a heart attack or stroke. Preventing or delaying another event could affect survival, disability, hospitalization and long-term care, although this secondary analysis does not separately establish reductions in each of those consequences.
The absolute numbers also deserve attention. In the primary SOUL results, the percentage of participants experiencing a first major cardiovascular event differed by 1.8 percentage points, 12.0% with semaglutide versus 13.8% with placebo, over a median follow-up of approximately four years.
The 108-event difference from the newer analysis measures something different. It counts events rather than unique people, because a single participant could contribute more than one event. Dividing that difference by the full study population would not produce a valid individual treatment-benefit estimate.
The total-event analysis strengthens the cardiovascular case for oral semaglutide. It does not establish a separate treatment effect on second and third events.
Any clinical benefit also has to be considered alongside tolerability. As previously reported from SOUL, adverse events led 15.5% of participants assigned to oral semaglutide to discontinue treatment, compared with 11.6% assigned to placebo. The secondary cardiovascular analysis does not replace a comprehensive evaluation of the drug’s safety.
The Limits the Authors Flag
The researchers are explicit about the limitations. Although the total-event analysis was prespecified, SOUL was not designed specifically to test treatment differences in subsequent cardiovascular events. The authors regard these analyses as exploratory, and the statistical findings were not adjusted for multiple comparisons.
That matters because the more comparisons researchers examine, the greater the possibility of finding an apparently significant result by chance. A prespecified analysis is generally more credible than one devised only after seeing the data, but prespecification does not turn a secondary, exploratory outcome into an independently confirmed treatment effect.
There are practical limitations too. The analysis followed the intention-to-treat principle, meaning participants remained in their assigned groups regardless of whether they continued taking the study medication. It therefore cannot determine whether subsequent events occurred while participants were actively receiving oral semaglutide.
The researchers also lacked detailed information about other cardiovascular medications taken between first and subsequent events. After a heart attack or stroke, treatment frequently changes. Those changes can affect the likelihood of another event, making subsequent-event comparisons harder to interpret even when the original treatment assignment was randomized.
SOUL was funded by Novo Nordisk, the manufacturer of oral semaglutide. The company shared responsibility for trial design and conduct with an academic steering committee, maintained the clinical database and performed the statistical analyses. Several authors were Novo Nordisk employees or shareholders. These relationships are disclosed in the published paper and are relevant context for readers evaluating the evidence.
The study population also limits generalization. SOUL enrolled people with type 2 diabetes and established cardiovascular disease, chronic kidney disease, or both. Its findings cannot automatically be extended to people without diabetes, to patients at substantially lower cardiovascular risk, or to semaglutide formulations tested at different doses.
Why Total Events Matter in the GLP-1 Market
As the GLP-1 market expands beyond weekly injections, cardiovascular evidence is becoming an increasingly important point of comparison. Novo Nordisk’s oral semaglutide products now compete in a market that includes Lilly’s orforglipron and other investigational oral GLP-1 receptor agonists. Grey Pepper has examined that competition in its coverage of Wegovy pill and Foundayo market-share data and its comparison of oral GLP-1 treatments.
Weight loss and blood-glucose control matter, but cardiovascular outcomes can influence prescribing guidelines, labeling and reimbursement decisions. A medicine that demonstrates fewer heart attacks, strokes or cardiovascular deaths may have a different clinical value from one supported primarily by weight-loss outcomes.
That does not mean the SOUL secondary analysis proves oral semaglutide is superior to other GLP-1 medicines. The trial compared oral semaglutide with placebo on top of standard care, not with Lilly’s orforglipron, injectable tirzepatide or another active GLP-1 treatment. Differences between clinical trials make direct numerical comparisons unreliable without appropriate comparative evidence.
The more interesting lesson is about measurement. Time-to-first-event analyses remain central to cardiovascular outcomes trials, but they cannot capture all the clinical burden accumulated by patients with recurring disease. Total-event analyses can add useful information, particularly for medicines expected to be taken over many years. Their interpretation, however, depends on the trial design and the statistical certainty of the findings.
What the SOUL Analysis Actually Establishes
For adults with type 2 diabetes and cardiovascular or kidney disease similar to those enrolled in SOUL, the evidence for oral semaglutide’s reduction of first major cardiovascular events is supported by a randomized clinical trial. The new analysis adds a statistically significant finding for the combined burden of first and subsequent events, with 683 total events in the semaglutide group against 791 on placebo.
The repeat-event finding is narrower. Subsequent events were numerically fewer with semaglutide, but the separate estimates for second and third events were not statistically conclusive. The investigators had not designed SOUL primarily to prove those individual effects, and the limited number of recurrent events makes that uncertainty important.
For patients, this is useful additional evidence about cardiovascular outcomes, but it is not a reason to change treatment independently of a clinician. Decisions still depend on an individual’s diabetes management, cardiovascular risk, contraindications, tolerability and other therapies.
For researchers and drugmakers, the lesson is that a completed trial can contain more information than its original primary endpoint reveals. Counting first events tells one part of the story. Counting every event helps describe the burden of disease over time. Proving that a medicine prevents repeat events specifically may require more targeted evidence.
The SOUL data show fewer cardiovascular events overall with oral semaglutide. They do not yet settle how much protection the pill offers after a patient has already experienced one.
Sources
- Marx N, Deanfield JE, Mulvagh SL, et al. First and Subsequent Major Cardiovascular Events With Oral Semaglutide in High-Risk Type 2 Diabetes: A Secondary Analysis of the SOUL Randomized Clinical Trial. JAMA Cardiology. Published online November 11, 2026. doi:10.1001/jamacardio.2026.4944.
- Oral Semaglutide and Cardiovascular Outcomes in High-Risk Type 2 Diabetes. New England Journal of Medicine. Published March 29, 2025. doi:10.1056/NEJMoa2501006.
- ClinicalTrials.gov — SOUL: A Heart Disease Study of Semaglutide in Patients With Type 2 Diabetes. Clinical trial registration NCT03914326.

