The honest answer depends on a distinction most summaries skip. The headline cardiovascular results are composites, and a composite moving is not the same as stroke moving. This site reads effect estimates the same way in sixteen percent lower, six hundredths of a point.
What the big trials actually counted
SELECT randomized 17,604 people with cardiovascular disease and a BMI of 27 or more, without diabetes, to semaglutide 2.4 mg or placebo [1]. A primary event occurred in 6.5% of the semaglutide group against 8.0% on placebo, a hazard ratio of 0.80 with an interval from 0.72 to 0.90.
That primary endpoint is death from cardiovascular causes, non-fatal myocardial infarction, or non-fatal stroke, counted as time to first event. Stroke is one of three components, and the trial was not designed to size stroke alone.
A meta-analysis of 11 trials covering 85,373 participants reports the same shape [2]. In the 67,769 participants with type 2 diabetes, GLP-1 drugs reduced MACE by 13% (HR 0.87, 95% CI 0.81 to 0.93). All-cause death fell 12% (HR 0.88, 0.83 to 0.93). The kidney half of that analysis sits alongside the growing dialysis evidence base.
The one trial built around stroke
LAMP randomized 636 patients with type 2 diabetes who had a minor ischemic stroke or a high-risk transient ischemic attack, within 24 hours of symptom onset [3]. One arm received liraglutide for 90 days on top of guideline care; the other received guideline care alone.
| Outcome at 90 days | Liraglutide + standard care | Standard care alone |
|---|---|---|
| Stroke recurrence | 25 of 316 (7.9%) | 44 of 318 (13.8%) |
| Excellent functional outcome | 87.3% | 77.8% |
Both outcomes favored liraglutide. The authors then state that, given the underpowered nature of the study, the findings should be interpreted with caution — a caveat this site usually has to supply itself.
Underpowered means the trial enrolled fewer patients than needed to reliably detect the effect it tested. A positive result from such a trial is more likely to overstate the effect, and more likely to shrink on replication.
Two further limits. The design was open-label with blinded endpoint assessment, so patients and treating doctors knew the assignment while outcome assessors did not. And participants were Chinese, which the authors note when framing their conclusion.
Where that leaves the question
Composite cardiovascular outcomes improve, consistently, across large trials and a large meta-analysis. Stroke measured on its own has one small dedicated trial that its own authors discount, and one large cohort in which it did not move.
So the defensible statement is that these drugs reduce cardiovascular events overall, and that the stroke-specific contribution is not established. That is a narrower claim than most pages make, and it is the one the evidence supports. The distinction between preventing something and treating it is set out in why preventing something is not treating it.
Nothing here is a reason to start or stop a medication after a stroke. That decision involves blood pressure, anticoagulation, the cause of the stroke and much else, and it belongs to the clinicians managing it.