The SELECT trial, published in The New England Journal of Medicine in November 2023, produced the headline that changed how the FDA labels semaglutide: a 20% relative reduction in major adverse cardiovascular events compared to placebo. That number is real. The question worth asking is who it applies to — because the enrollment criteria in SELECT exclude the majority of people currently being prescribed Wegovy.
My argument is not that the SELECT findings are inflated. They are well-conducted and genuinely important. My argument is that the trial answered a precise question about a specific population, and both the press coverage and the downstream prescribing behavior have treated that answer as more general than it is. Understanding the gap is not academic. It changes what the drug is actually doing for whom.
Let me go through the trial design first, because the design is the argument.
Who was in SELECT — and who wasn't
SELECT enrolled 17,604 adults across 41 countries. The inclusion criteria were: age 45 or older, BMI of 27 or higher, and — critically — established cardiovascular disease, defined as prior myocardial infarction, prior stroke, or symptomatic peripheral arterial disease. The exclusion criteria included diabetes at baseline.
Read that carefully. Everyone in SELECT already had a heart attack, a stroke, or vascular disease severe enough to be symptomatic. Nobody had diabetes. This is a high-risk secondary-prevention population, not a general obesity treatment population. The median age at enrollment was 61.6. The average BMI was 33.4. Over 73 percent were male.
Now consider who is actually receiving Wegovy prescriptions in 2024 and 2025. Commercial claims analyses consistently show the median prescribing age in the low-to-mid 40s, majority female, significant proportion without established CVD. That person — the 42-year-old woman with obesity and hypertension but no prior cardiac event — was not in SELECT. Applying the SELECT cardiovascular risk-reduction findings to her prescription decision requires an extrapolation the trial data does not support.
The 20% number: relative vs. absolute
The primary outcome in SELECT was a composite of cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke — what researchers call three-point MACE. Over a median follow-up of 39.8 months, the event rate was 6.5% in the semaglutide group and 8.0% in the placebo group.
The relative risk reduction is 20%. That is the number that traveled. The absolute risk reduction is 1.5 percentage points. The number needed to treat to prevent one MACE event over roughly 3.3 years is approximately 67.
None of these framings is more "correct" than the others — they describe different aspects of the same result. But the relative framing dominates coverage because it is larger and simpler. For a patient trying to understand whether this trial is relevant to their situation, the NNT of 67 over 3.3 years in a population with established CVD is the most useful number. It is also the one you will not find in most media summaries.
The weight-independence question
The most scientifically interesting debate SELECT opened is whether the cardiovascular benefit is partially independent of weight loss. In STEP-1, the matched population (similar drug, similar dose, no required CVD history) lost roughly 14 to 17 percent of body weight. In SELECT, the weight loss was lower — about 9.4 percent at two years. Yet the cardiovascular outcomes diverged significantly.
Some researchers, notably Lincoff and colleagues in a subgroup analysis published alongside the main trial, argued that the magnitude of cardiovascular benefit exceeded what weight-loss models would predict based on the observed weight change alone. They pointed to plausible mechanisms: semaglutide's effects on inflammation (CRP reduction was significant in SELECT), natriuretic effects that reduce blood pressure, and direct cardiac signaling via GLP-1 receptors in heart tissue.
If SELECT's benefit is partly weight-independent, then the drug is doing something in the cardiovascular system that we do not yet fully understand — which is a reason for more research, not a reason to generalize the finding to populations where it was not tested.
The counter-argument, which I find more compelling, is that disentangling weight loss from other mechanisms in a trial where both are occurring simultaneously is methodologically difficult. The subgroup analyses are hypothesis-generating, not confirmatory. The observed 9.4% weight loss in SELECT, spread over a population with significant comorbidities and real-world adherence patterns, is still meaningful — and it may account for more of the cardiovascular benefit than the optimistic weight-independent interpretation suggests. We do not know yet.
What the FDA approval actually says
In March 2024, the FDA expanded Wegovy's indication to include reduction of cardiovascular death, non-fatal MI, and non-fatal stroke "in adults with established cardiovascular disease and either obesity or overweight." The label language mirrors the enrollment criteria almost exactly. Established CVD is required. This was a deliberate and defensible regulatory decision — the agency approved what the trial showed, for the population the trial tested.
What has happened in practice is that the approved indication has been used to justify prescribing in a much broader population. Some of that prescribing is defensible on grounds outside SELECT — weight loss itself has cardiovascular benefits supported by other evidence, and treating obesity has value beyond any single trial. But it is a different argument, built on different evidence, from what SELECT provides.
The SELECT numbers in context
For a population of 1,000 high-risk patients matching SELECT criteria — established CVD, non-diabetic, BMI ≥27, median age 61 — treating all of them with semaglutide for 3.3 years would be expected to prevent roughly 15 MACE events compared to placebo. That is a clinically meaningful number in a population where a cardiac event carries serious mortality risk. It is also a number that requires long, continuous, expensive treatment with a drug that 23 percent of real-world patients discontinue within 12 months.
Combining SELECT's outcomes data with real-world discontinuation rates produces a more sobering picture than either number alone. The trial's benefit was measured in patients who stayed on the drug through a median of 39.8 months. If real-world persistence tracks the commercial claims data — roughly 25 percent still filling at 12 months — the population-level cardiovascular benefit will be substantially smaller than SELECT's controlled conditions imply.
What SELECT actually means
SELECT is a landmark trial. It established, for the first time in a large randomized controlled trial, that a GLP-1 receptor agonist reduces hard cardiovascular endpoints in non-diabetic patients with obesity and established CVD. That is a genuinely new result, and it is the kind of result that matters for cardiology practice in a specific, high-risk group.
What it does not do is validate cardiovascular risk reduction as a general justification for prescribing semaglutide to people who do not match the SELECT population. That extrapolation may turn out to be correct — lower-risk populations might benefit too, and ongoing trials are designed to test that. But as of mid-2026, SELECT tells us what happens in one specific, well-defined group. Treating the 20% figure as a general property of the drug, rather than a finding in that group, is a category error. It is also the error the headlines made, repeatedly, when the trial published.
Ozemback — June 2026