A study made the rounds recently claiming semaglutide cuts the risk of heart attack, stroke, or death by more than half compared with tirzepatide. The number is real. It comes from a published, peer-reviewed paper with over 21,000 patients in it. It is also a study that Novo Nordisk — the company that makes semaglutide and sells it as Wegovy — designed, funded, and staffed with its own employees as every listed author.
That doesn’t make the finding false. It does mean the number needs a different kind of reading than the headline gives it.
What the study found
The study is called STEER, published in Diabetes, Obesity and Metabolism. It’s a retrospective analysis of a US insurance claims database (Komodo Research Data), looking at 21,250 adults, 45 and older, with overweight or obesity and an existing history of heart attack, stroke, or peripheral artery disease — no diabetes. Each patient had started either semaglutide or tirzepatide, and the two groups were matched using propensity scores to balance age, sex, and other baseline characteristics.
The results, comparing semaglutide against tirzepatide:
- Intent-to-treat analysis: a 29% lower risk of a 3-point cardiac event composite (heart attack, stroke, death) and a 22% lower risk of a 5-point composite that adds heart failure hospitalization and coronary procedures.
- Per-protocol analysis (patients censored once they stopped filling their prescription): a 57% lower risk of the 3-point composite and a 43% lower risk of the 5-point composite.
Those are large, statistically significant differences between two drugs that both get marketed as best-in-class for people with obesity and heart disease.
What “real-world” and “retrospective” actually mean here
STEER is not a clinical trial. Nobody was randomly assigned to take semaglutide or tirzepatide. Researchers instead pulled insurance claims for people who were already prescribed one drug or the other and compared what happened to them afterward.
Propensity matching narrows the gap between two such groups, but it can only balance the characteristics the researchers thought to measure and that show up in claims data — age, sex, recorded diagnoses. It can’t fully correct for why a clinician chose one drug over the other for a specific patient, which health system they were in, how consistently they filled either prescription, or dozens of other factors that never make it into an insurance claim. That gap is called confounding by indication, and it’s the standard limitation of every observational drug comparison, not a flaw unique to this one.
The follow-up window matters too. Patients were tracked for a matter of months, not years — short enough that a handful of additional events in either group can swing a “percent reduction” a long way. A 57% relative risk reduction sounds enormous until you know how few total events it’s built from.
Who ran it
Every listed author on the STEER paper — seven people — is disclosed in the paper itself as an “employee and shareholder of Novo Nordisk at the time of the study.” The copyright line reads ”© 2026 Novo Nordisk Inc.” The data was pulled, matched, analyzed, and written up entirely inside the company that sells the drug the study says wins.
That’s disclosed, not hidden — which is exactly why it’s checkable, and exactly why it belongs in how you weigh the finding. An industry-funded study isn’t automatically wrong. But a company running a head-to-head analysis of its own drug against a competitor’s, using a study design that can’t fully rule out confounding, is not the same evidentiary weight as an independent randomized trial. It’s closer to a well-produced piece of competitive marketing that happens to have been peer reviewed.
For comparison: semaglutide’s actual randomized evidence in this population is the SELECT trial — 17,604 patients with existing cardiovascular disease, randomly assigned to semaglutide or placebo and followed for a mean of nearly 40 months. SELECT found a real, statistically significant 20% reduction in major cardiac events against placebo. Notably, SELECT was also funded by Novo Nordisk — but random assignment and multi-year, blinded follow-up remove the specific problem STEER has. Funding source matters less when the study design itself controls for who ends up in which group. Tirzepatide doesn’t yet have an equivalent head-to-head randomized trial against placebo in this exact population, which is part of why a claims-based comparison like STEER exists in the first place — it fills a gap real trials haven’t yet, using a method that can’t fully close it.
What this has to do with compounded medications
Every number above describes the manufactured, brand-name products — Wegovy and Zepbound — dosed exactly as labeled. Nothing here says anything about compounded semaglutide or tirzepatide, the versions prepared individually by a licensed pharmacy rather than made by the original manufacturer. Compounded medications haven’t been through SELECT, STEER, or any comparable study, and findings about the branded product don’t transfer to a compounded preparation. Whether either drug — compounded or branded — makes sense for you is a question for a licensed clinician who can review your cardiac history, not a percentage from a claims database.
The actual lesson
The honest takeaway from STEER isn’t “semaglutide wins.” It’s that a lot of the comparative claims circulating about GLP-1 medications — including the flattering ones — come from the companies selling them, using real-world data that’s genuinely useful for spotting patterns but genuinely limited in what it can prove about cause and effect. That distinction gets lost in a press release, and it gets lost faster in an ad. Reading past it is the whole point of looking at how these programs are priced and structured instead of taking a marketing claim at face value — the medication that’s right for you is a clinical decision, not a leaderboard.
This post is general information, not medical advice. It does not describe a benefit of any NoBsRx program, and it is not a claim that one medication is safer or more effective than another for any individual. Compounded medications are not FDA-approved. Treatment is never guaranteed — eligibility, diagnosis, and every treatment decision are made by independent licensed providers based on an individual clinical assessment, and availability varies by state. Do not start, stop, or switch any medication based on an article. If you have a history of cardiovascular disease, talk to a licensed clinician before starting a GLP-1 medication.