Cardiovascular outcomes: what SELECT showed

Why this one trial keeps coming up

If you have noticed clinicians talking about these medicines differently over the last couple of years, SELECT is a large part of the reason. Before it, the argument for semaglutide outside diabetes rested on weight-change trials and on the conditions that travel with excess weight. SELECT asked a harder question: does the drug prevent heart attacks, strokes and cardiovascular deaths in people who do not have diabetes?

That is a different sort of trial and a much more expensive one. You cannot answer it in 68 weeks with a scale. You need many thousands of people, several years, and events you count rather than measurements you take.

Sunil, who runs our research reading group, spent three sessions on this paper and his summary is worth borrowing: SELECT is a genuinely good trial that answers a genuinely narrow question, and almost every argument you will see online about it consists of stretching that narrow answer somewhere it was never tested.

Who was actually in it

This is always the first question in our reading group, and with SELECT it does most of the work.

SELECT (Lincoff, New England Journal of Medicine, 2023) enrolled 17,604 people. To get in, you had to be at least 45 years old, have a body mass index of 27 or above, and have established cardiovascular disease — a previous heart attack, a previous stroke, or symptomatic peripheral arterial disease. And you had to not have diabetes.

Read that list again, because it defines the ceiling of what the trial can tell anybody. This is a secondary prevention population: people who have already had a cardiovascular event and are therefore at high risk of another. They were older than the average person in this community. They were, on the whole, already on statins and blood pressure treatment and antiplatelets, because that is standard care after an event — so the drug was tested on top of good treatment, not instead of it. Roughly a quarter were women, which is a real limitation of the trial and one the authors acknowledge.

Participants received semaglutide 2.4 mg weekly or placebo, and were followed for a mean of a little over three years.

What it found

The primary endpoint was a composite: death from cardiovascular causes, non-fatal heart attack, or non-fatal stroke. Composites are standard in this kind of trial because single events are too rare to power a study on, but they always deserve a look inside.

The result: 6.5 per cent of the semaglutide group had one of those events, against 8.0 per cent of the placebo group. Expressed as a hazard ratio, that is 0.80, with a 95 per cent confidence interval of 0.72 to 0.90 — a roughly 20 per cent relative reduction, and the confidence interval sits comfortably away from 1, so this is not a marginal finding.

The two framings of the same number are both true and it is worth holding both. A 20 per cent relative reduction sounds transformative. An absolute difference of about 1.5 percentage points over three years, in a high-risk population, sounds more modest. Both describe the identical result. The relative figure travels better in headlines; the absolute figure is the one that tells you what it might mean for a person.

Deaths from cardiovascular causes and from any cause both trended in favour of semaglutide. Participants also lost weight — this was, after all, semaglutide 2.4 mg — but the trial was not designed to answer whether the cardiovascular benefit was caused by that weight change.

The mechanism question nobody can answer yet

Here is the bit our reading group enjoyed most, because it is genuinely unresolved.

The curves separated early — earlier, several commentators argued, than a benefit driven purely by weight change would plausibly appear. Blood pressure, inflammatory markers and lipids all shifted in the trial, and there is decent laboratory evidence for direct effects on vascular inflammation. So there are several candidate explanations: weight change, glycaemic effects short of diabetes, blood pressure, inflammation, a direct vascular action, or some combination.

SELECT cannot distinguish between them. It was not designed to. Mediation analyses have been published and they are interesting, but a mediation analysis is a statistical argument about an observed association, not an experiment.

Sunil’s standing objection is to the sentence "it works through inflammation, not weight", which he has now seen asserted with total confidence by people on both sides of that argument. What the trial establishes is that the events happened less often. Why is a live scientific question and anybody claiming otherwise is selling something.

What it does not show

Every one of these is a claim we have seen made about SELECT that the trial does not support.

• That the drug prevents heart attacks in people without existing cardiovascular disease. Nobody like that was enrolled. It may; it has not been tested.
• That it works as well in a younger population. The minimum age was 45 and the mean was considerably higher.
• That the benefit applies to other GLP-1 medicines automatically. Drugs in a class are not interchangeable for outcomes. LEADER (Marso, NEJM, 2016) found a reduction in major cardiovascular events with liraglutide in people with type 2 diabetes, and SUSTAIN-6 found a similar direction for injectable semaglutide in that population — but tirzepatide has no completed equivalent cardiovascular outcome trial in this setting, and absence of a trial is not evidence of absence of benefit or of harm.
• That it applies to compounded or research-sourced material. SELECT tested a manufactured product with known contents at a known dose. What is in an unregulated vial is a separate question entirely — see what purity actually means.
• That anyone should stop their statin. The trial was conducted on top of standard cardiovascular treatment, not instead of it. This one genuinely worries us when we see it in the wild.

What it might mean for you

We are not going to tell you. What we can say is what has changed in the conversations members report having.

People with established cardiovascular disease and no diabetes now find that some clinicians frame this medicine as cardiovascular risk reduction rather than as weight management, and in several countries that framing has affected what is licensed and what is funded. Mark, in our over-60 circle, said the shift in his cardiologist’s language — from something optional to something worth trying — was the reason he stopped feeling embarrassed about the prescription.

If you are in that group, the useful questions are: does this trial population resemble me, what would we be hoping for, and what would make us stop? Those fit in a short appointment. Talking to your GP when time is short has the wording that has worked for members.

And if you are not in that group — younger, no cardiac history — SELECT is interesting context and nothing more. It is not a promise made to you. Sunil would want that in bold.