Semaglutide cuts heart events by a fifth in high-risk people, trials confirm
Semaglutide cuts major heart events by about a fifth in high-risk people with heart disease, but the size swings across studies and the full safety story isn't locked down yet.
What does clinical evidence show about the relationship between GLP-1 receptor agonist drugs (such as Ozempic/semaglutide) and cardiovascular disease risk?
- 1The SELECT trial followed 17,604 high-risk people and found semaglutide cut major heart events by about a fifth versus placebo.
- 2Independent pooled reviews land anywhere from 13% to 32% fewer heart events because they pool different drugs, patients, and study types.
- 3Not every GLP-1 drug helps the heart — two big trials, ELIXA and EXSCEL, came up flat.
- 4Europe's drug regulator flagged a very rare vision-loss side effect, about one extra case per 10,000 people treated for a year.
- 5Most of the heart benefit seems to come from something other than the pounds lost, but that mechanism number rests on a single study.
Millions inject semaglutide hoping it guards their heart. The biggest trial, SELECT, followed 17,604 high-risk people and found the drug cut major heart events by about a fifth versus a dummy shot. Two other big studies backed it. All three focused on folks who already had heart disease, diabetes, or both. Real-world data went even higher — some studies claimed cuts of 45% — but those numbers come with strings attached: shorter tracking time, industry money in the mix, and study designs that can inflate results. Not every drug in the same family works. Two big trials, ELIXA and EXSCEL, came up flat. So the heart benefit is not automatic across the whole group. Pooled reviews swing from 13% to 32% fewer events because they pool different studies and patient types. Most of the heart protection seems to come from something other than just shedding pounds — one analysis pinned roughly 80% of it on non-weight factors — but that figure rests on a single study and hasn't been replicated yet. Safety isn't alarming, but it isn't closed either. Europe's drug regulator flagged a very rare vision-loss risk — about one extra case per 10,000 people treated for a year. In March 2026 the FDA cited the maker for serious violations over reporting three patient deaths late, though the FDA stressed the letter does not mean the drug caused them. The single biggest trial did not show fewer heart deaths on its own — the win rides on the combined score of death, heart attack, and stroke together. Bottom line for high-risk people with established heart disease: a genuine, meaningful heart benefit backed by solid trials, wrapped in real uncertainty about how big it really is, how it works, and whether the full long-term safety picture has been independently confirmed yet.
The Full Investigation
7 sections · 7 min read
Sensitive topic — how we handled it
This is an elevated-sensitivity topic (such as pandemic policy, casualty figures, migration, or an identity-charged subject). Inquesta applies heightened sourcing standards here: contested figures are presented with both sides, single-source claims are flagged, and the full methodology and source list are shown in place of a summary. Read contested numbers with extra care.
Why everyone suddenly cares whether Ozempic is good for your heart
Semaglutide is now widely prescribed -- sold as Ozempic and Wegovy -- and many patients and doctors are considering whether it guards the heart, not just the waistline. That question has real stakes. The people most likely to take these drugs already carry heavy heart-disease risk.
The question grew from a landmark study. On November 11, 2023, Dr. Abraham Lincoff presented the SELECT trial at a big American Heart Association meeting. The trial was paid for by Novo Nordisk, the drug's maker. Regulators moved too. The FDA approved Ozempic to cut heart-event risk in diabetics with heart disease back on January 16, 2020.
So here's the map. The evidence splits four ways: what the trials show, what the safety warnings say, what the big pooled reviews conclude, and how the drug might even work. Follow all four and you get the real answer.
The trials show heart events dropping by roughly a fifth
Start with the big one. SELECT enrolled 17,604 people aged 45 and up, all overweight and all with existing heart disease. Semaglutide cut major heart events -- heart death, heart attack, stroke -- by 20% versus a dummy shot (HR 0.80; 95% CI 0.72-0.90; P<0.001). That is a real, statistically solid result.
But dig one layer down. Heart death alone did not clear the bar. It came in at 2.5% on the drug versus 3.0% on placebo -- a gap that could be chance (HR 0.85; p=0.07). So the win rides on the combined score, not on fewer deaths by itself.
A second trial backs the direction. SUSTAIN-6 split 3,297 diabetics and found a 26% drop in three-point heart events, driven mostly by a 39% cut in non-fatal stroke. Two separate trials, overlapping results -- 20% and 26%. The gap likely reflects who was studied: SELECT took heart-disease patients, SUSTAIN-6 took diabetics.
One real-world study went much bigger. The SCORE study reported a 45% drop in five-point heart events over about 200 days. That's roughly double the trial numbers. Watch out: SCORE is observational, short, and the study disclosed author ties to Novo Nordisk. Short observational studies can inflate benefits, so treat that 45% as a ceiling, not a measurement.
Open: Does SELECT's non-significant cardiovascular-death result become significant with longer follow-up?; Can the SCORE real-world 45-57% reduction be replicated by an independent group?
Regulators flagged rare vision loss -- and cited the maker for late death reports
Benefit is only half the story. Regulators have been busy on the risk side, and this is where the drama lives. Europe's drug watchdog, the EMA, ruled that a rare form of sudden vision loss called NAION is a very rare side effect -- up to 1 in 10,000 people, a frequency category -- while epidemiological studies suggest about a two-fold higher risk, corresponding to roughly one additional case per 10,000 person-years of treatment.
In March 2026 the FDA sent Novo Nordisk a warning letter citing serious violations in safety reporting, including failing to report three patient deaths by legal deadlines. The FDA said plainly the letter does not mean semaglutide caused those deaths -- the problem was how fast the company shared the information.
The trials also flagged tolerability. In SELECT, people quit the drug twice as often as placebo -- 16.6% versus 8.2% -- mostly from gut side effects. And a huge database scan tells the other side. A review of over 6 million adverse-event reports found 2,398 involving oral semaglutide and zero major heart events, with a disproportionality measure (ROR 0.71) below the threshold that would flag a safety concern. No red flag there.
2020-01-16
- FDA approves Ozempic to reduce cardiovascular event risk in adults with type 2 diabetes and known heart disease
2023-11-11
- Dr. Abraham Lincoff presents SELECT trial results at American Heart Association Scientific Sessions
2024-03-08
- FDA approves Wegovy to reduce cardiovascular event risk in overweight and obese patients with established CVD
2024-09-12
- American College of Cardiology publishes SELECT trial summary showing 20% MACE reduction with semaglutide
2025
- EMA Pharmacovigilance Risk Assessment Committee concludes NAION is a very rare side effect of semaglutide
2026-03-10
- FDA issues warning letter to Novo Nordisk for 'serious violations' in semaglutide safety reporting, including failure to report three patient deaths by legal deadlines
Open: Is the EMA's NAION risk estimate confirmed by any independent epidemiologic study?; What did the FDA warning letter conclude about the underlying cause of the three reported deaths, if anything?
The big pooled reviews agree on direction but not on size
Zoom out from single trials and the picture gets messier. When researchers pool many studies, they all point the same way -- fewer heart events -- but the size swings wildly. That spread is the real story here.
Count them off. A 21-trial review of 99,599 patients found a 13% drop, rated high-certainty (IRR 0.87; NNT 66). A 9-trial review of 63,613 patients found 14% (HR 0.86). An observational-only review of over 500,000 (504,029) people found 28% (HR 0.72). A smaller 11-trial review found 32% (OR 0.68). Why the range? The credible trial-only reviews cluster near 13-14%; the observational and semaglutide-specific ones run higher because they pool different patients and study types, and odds ratios can overstate the effect.
Two of these reviews flagged their own weakness. The Cardiology Research authors admitted their funnel plot looked lopsided -- a hint of publication bias -- with just 5 studies. The Biomedical Reports team found a statistically significant asymmetry too (Egger's P=0.038), signaling missing negative studies. Others came out clean: the 9-trial review showed no such bias (Egger's p=0.154).
One claim to watch. A promotional site reported a 20% cut across 16 trials and 13-14% across six big trials. Those numbers fit the credible reviews, but the site has no named authors or traceable source -- so treat them as unconfirmed, not proof.
And the crucial caveat: not every drug in this class works. Two big trials, ELIXA and EXSCEL, came up flat on heart events. Benefit is not automatic across the whole GLP-1 family.
Open: What is the identifiable primary publication behind the PeptideJournal 16-trial and 6-trial figures?; How much does pooling neutral trials like ELIXA and EXSCEL dilute the class-wide effect estimate?
The benefit seems to come from more than just weight loss
If the drug works, how does it work? This is where scientists get genuinely curious -- and honest about what they don't know. The obvious guess is weight loss. The evidence says it's more complicated.
Mediation analyses -- statistical tools that try to split a benefit into its causes -- point past the scale. One SELECT analysis reported that about 33% of the heart protection tracked with shrinking waistlines (HR 0.86). Another found that roughly 80% of the heart benefit came from factors other than body weight. Read together, most of the payoff seems to run through something besides the pounds lost. Each of those numbers, though, rests on a single study.
Blood pressure works differently. Here, about 89% of the drop was tied to weight loss. The drugs shave systolic pressure by a modest 2-5 mm Hg across many patient types.
There's a wrinkle on heart rate. GLP-1 drugs nudge it up 2-5 beats per minute, which raises a theoretical worry for people with rhythm problems or blocked arteries. But the big trials found no extra heart events from that bump. A tiny 24-patient Japanese study offers a hint at the mechanism: blood-vessel function improved sharply, and that gain did not track with weight, blood sugar, or cholesterol changes. Small study, but it points the same direction -- benefits beyond weight.
Open: Would a trial comparing semaglutide against diet-matched weight loss confirm the weight-independent pathway?; Can the single-study mediation proportions (33% waist, 80% non-weight) be reconciled and replicated?
Testing the competing explanations
So what best explains the whole picture? Four readings compete, and the evidence sorts them unevenly.
One explanation says the heart benefit is mostly just weight loss and its knock-on effects -- smaller waist, better blood pressure and sugar. Some evidence fits: a third of the protection tracked waist size, and 89% of the blood-pressure drop tracked weight. But it runs into a wall. The SELECT analysis pinning about 80% of the heart benefit on non-weight factors directly cuts against it. This reading looks weak.
A second reading holds that the drug protects the heart through direct routes -- calming inflammation, healing blood-vessel lining -- separate from weight. The 80% non-weight figure supports it, and the Japanese vessel-function study points the same way. Nothing in the evidence flatly contradicts it. It's the more plausible story, though built on single studies.
A third reading claims every GLP-1 drug helps the heart equally. The pooled reviews tempt you toward it. But ELIXA and EXSCEL came up flat, which breaks the class-effect idea. Weak.
A fourth reading warns that rare serious risks -- vision loss, late-reported deaths -- could chip away at the net benefit, especially for lower-risk users. What pushes back: the FDA said the deaths weren't shown to be drug-caused, and the 6-million-report database found no heart signal. Plausible as a caution, not established as a harm.
Open: Would a head-to-head trial of different GLP-1 drugs settle whether benefit is a class effect or drug-specific?; Would benefit-harm modeling across risk strata show the rare risks materially offsetting benefit in lower-risk patients?
What the evidence forces us to conclude
Back to the bet millions are placing. Here's what the evidence can and can't promise them.
For people who already have heart disease plus obesity or diabetes, the direction of effect is solidly established: semaglutide cuts major heart events by roughly a fifth, and two independent trials agree. Nothing in this evidence base shows the drug raising heart-disease risk. The size of the benefit varies a lot across studies (13% to 32%), but every credible estimate points the same way -- fewer events, not more. The mechanism is only partly understood, and the 2026 FDA warning over late death-reporting keeps the full safety picture from being airtight. On the central question, the honest answer for that high-risk group: benefit, not harm, with real uncertainty about how big, how it works, and whether long-term safety is fully known.
But the caveats are real and load-bearing. The single biggest trial did not show fewer heart deaths on its own. The pooled reviews swing from 13% to 32%, and the higher numbers lean on weaker methods. Not every drug in the class delivers. And the mechanism story, while pointing beyond weight loss, rests on single studies that haven't been replicated.
The safety file stays open, not alarming. The rare vision-loss risk is quantified and small. The 2026 FDA warning was about reporting speed, not demonstrated harm -- but it does mean the full safety picture hasn't been independently confirmed. Weighing it all: a genuine, meaningful heart benefit in high-risk patients, wrapped in real uncertainty about size, mechanism, and long-term safety. Premature to call it settled; wrong to call it harmful.
Why it matters
These drugs are now everyday medicine for millions with heart risk, and the FDA has approved them specifically to protect the heart. Getting the answer right decides who benefits, who faces a rare vision risk, and whether patients and doctors can trust that the full safety story has been shared on time. The stakes are personal: real protection for high-risk people, real uncertainty everywhere else.
- Whether the heart benefit holds up in lower-risk people without established cardiovascular disease, since the pivotal trials enrolled high-risk patients.
- The long-term (multi-year) cardiovascular and safety trajectory, since key real-world data cover only about 200 days.
- How much of the evidence base is shaped by Novo Nordisk sponsorship of the pivotal trials and industry-affiliated authorship of key real-world studies.