Semaglutide and heart risk: how to read SELECT’s 20% result

The question

SELECT found fewer major cardiovascular events with semaglutide than with placebo in people who already had cardiovascular disease and overweight or obesity, without diabetes. The often-quoted 20% reduction describes a relative time-to-event comparison. It does not mean that 20 out of every 100 treated people avoided an event.

The randomized trial included 17,604 participants. A primary event occurred in 569 of 8,803 semaglutide participants (6.5%) and 701 of 8,801 placebo participants (8.0%). The hazard ratio was 0.80, with a 95% confidence interval of 0.72–0.90. Mean follow-up was 39.8 months. The primary outcome combined cardiovascular death, nonfatal heart attack and nonfatal stroke. Original SELECT report.

Keep the relative and absolute views together

Subtracting the rounded event proportions gives 1.5 percentage points: 8.0 minus 6.5. That helps show the scale of the observed difference. It is not the same calculation as the reported hazard ratio, which incorporates the timing of events during follow-up.

This is why a headline can accurately mention a 20% relative reduction while the event proportions differ by 1.5 percentage points. The two numbers answer related but different questions. Removing the denominator or the observation period makes either one easier to misread.

A purely hypothetical example makes the distinction easier to see. If a risk fell from 10% to 8%, the absolute difference would be two percentage points and the relative reduction would be 20%. If it fell from 1% to 0.8%, the same relative reduction would correspond to 0.2 percentage points. These are teaching examples, not SELECT estimates or individual forecasts.

Avoid treating the rounded SELECT proportions as a permanent lifetime risk estimate. A study observes people over a period; it does not tell us their entire future. A personal estimate also depends on how closely someone resembles the trial population and on their wider medical care.

A combined outcome is not three separate promises

The primary outcome counted a first occurrence of any of three specified events. An improvement in that combined outcome does not automatically establish an identical percentage reduction for each component. SELECT outcome definition.

When a summary switches from “major cardiovascular events” to “heart attacks,” it may have narrowed the endpoint without updating the evidence. Likewise, a statement about cardiovascular death is not interchangeable with a statement about all causes of death.

The practical reading habit is simple: keep the original outcome label beside the number. If a writer wants to discuss an individual component, look for that component's separate result and uncertainty. One impressive headline statistic should not be recycled across several different claims.

Who was the result about?

Existing cardiovascular disease is central to the study question. “Without diabetes” does not mean “healthy volunteers,” and “overweight or obesity” does not erase the participants' cardiovascular history. The trial was funded by Novo Nordisk. SELECT population and funding.

This matters when someone asks whether the finding applies to a person with no established cardiovascular disease. The study cannot simply be relabeled as a prevention trial in that different group. The appropriate response is to look for evidence addressing that population, rather than assume that the same relative and absolute effects transfer unchanged.

It also matters when reading a before-and-after story. SELECT compared randomized groups; an individual story about feeling better cannot reproduce that comparison. Stories can raise useful questions, but they do not supply the missing control group.

What the FDA indication adds

In March 2024, FDA approved Wegovy injection to reduce the risk of cardiovascular death, heart attack and stroke in adults with cardiovascular disease and either obesity or overweight. The agency specified use alongside a reduced-calorie diet and increased physical activity. Its announcement also discusses warnings and contraindications. FDA approval announcement.

An indication identifies a medicine and a defined use. It is not approval of every semaglutide preparation offered online. Nor does an average benefit eliminate the need to assess adverse effects and suitability with a clinician. Our compounded semaglutide guide explains why the ingredient name and finished product are separate questions.

For another outcome-specific reading exercise, see semaglutide and the FLOW kidney trial. Keeping the population, endpoint, comparison and time period together makes both studies easier to understand without turning either into a universal promise.