What You'll Learn
Table of Contents
What SELECT Actually Tested
SELECT was a multicentre, double-blind, randomised, placebo-controlled, event-driven superiority trial. Every one of those words is doing work, and this is about as strong as a design gets outside of the very largest cardiology programmes.
17,604 people were randomly assigned to weekly semaglutide 2.4 mg or placebo and followed for a mean of 39.8 months — a little over three years. The primary endpoint was a composite: death from cardiovascular causes, non-fatal heart attack, or non-fatal stroke, whichever came first.
That composite matters for how you read the headline. The trial was designed and powered to detect a difference in all three combined. It is not the same claim as “the drug prevents cardiovascular death,” and reporting it that way — which is common, and which an earlier version of our own metabolic health page did — overstates what was demonstrated.
Bottom line: A composite endpoint is a legitimate and standard design. It is also the point at which a result gets simplified in translation. When you see a cardiovascular trial summarised in one sentence, the first question worth asking is which events were being counted.
The 20% And The 1.5 Points Are The Same Number
Here are the raw counts. A primary endpoint event occurred in 569 of 8,803 people on semaglutide (6.5%) and 701 of 8,801 on placebo (8.0%). The hazard ratio was 0.80, 95% confidence interval 0.72 to 0.90, p<0.001.
Both of the following sentences describe that result correctly:
- “A 20% reduction in major cardiovascular events.” That is the relative risk reduction. It is true.
- “An absolute reduction of 1.5 percentage points over roughly three years.” That is the absolute risk reduction. It is the same finding, and it is also true.
Translated into the form clinicians tend to use: roughly 67 people in this population would need to be treated for about three years to prevent one of those events. That is a genuinely worthwhile number for a secondary-prevention drug. It is not the number most people picture when they read “20%.”
Neither framing is dishonest. But only one of them is used in marketing, and it is not the one with the percentage points in it.
Who Was Actually In The Trial
This is the part that decides whether the result applies to any given reader, and it is almost always dropped from the summary. To enter SELECT you had to be:
- 45 years of age or older
- With pre-existing cardiovascular disease — established, documented, already there
- With a BMI of 27 or greater
- And with no history of diabetes
So this is a secondary-prevention trial. Everybody in it had already had a cardiovascular event or established disease, which is precisely why their event rate was high enough — 8% over three years in the placebo group — for a difference to be detectable at all.
A 38-year-old with a BMI of 29, no cardiac history and no diabetes was not represented in this trial. Their three-year baseline risk of heart attack, stroke or cardiovascular death is a small fraction of 8%. Applying a 20% relative reduction to a much smaller baseline gives a much smaller absolute benefit, and SELECT provides no direct evidence about that group either way.
That is not a criticism of the trial. It is what the trial was designed to answer. The error is in the extrapolation, not the study.
Bottom line: Relative risk reductions travel between populations more readily than absolute ones. The same 20% applied to a low-risk person produces a benefit so small it can be outweighed by side effects — which is the whole reason risk-based prescribing exists.
Was It The Weight Loss, Or The Drug?
Participants on semaglutide lost a mean of 10.2% of body weight against 1.5% on placebo, with waist circumference down 7.7 cm against 1.3 cm. Substantial weight loss on its own reduces cardiovascular risk factors, so the obvious question is whether the drug did anything the weight loss would not have done.
SELECT cannot answer this. There was no arm that lost the same weight by other means. The trial compared drug against placebo, so weight loss and drug exposure are entangled by design — that is a limitation of the question, not a flaw in the execution.
Plausible contributors beyond weight include blood pressure reduction, improved glycaemic measures, and effects on inflammatory markers. Some of these have been examined in secondary analyses. None of it establishes a mechanism independent of weight loss, and anyone telling you the cardiovascular benefit is definitively “not just the weight” is stating a hypothesis as a finding.
For most practical purposes this distinction changes nothing: the benefit was observed with the drug, in that population, at that dose. It matters mainly if you are trying to reason about whether the effect persists after stopping — and SELECT does not address that either.
The Number On The Other Side Of The Ledger
The same paper reports something the headlines rarely carry. Adverse events leading to permanent discontinuation of the trial product occurred in 1,461 patients (16.6%) on semaglutide against 718 (8.2%) on placebo, p<0.001.
Twice the rate. One in six people randomised to the drug stopped taking it because of side effects, in a controlled trial with monitoring and support, over roughly three years.
Put beside the benefit figure, that is the honest trade-off: about 1.5 people per 100 avoided a major cardiovascular event, and about 8 more per 100 stopped the drug because they could not tolerate it. Both numbers come from the same table of the same paper. Only one of them tends to get quoted.
This is not an argument against the drug. For a 60-year-old who has already had a heart attack, that trade may be clearly worth making, and that is a conversation for their cardiologist. It is an argument against reading the benefit in isolation.
Adverse events leading to permanent discontinuation occurred in 16.6% of the semaglutide group against 8.2% on placebo (p<0.001), over a mean 39.8 months of exposure in 17,604 participants. Reported in the same paper as the benefit, and from the same population.
PMID: 37952131 ↗What SELECT Does Not Show
Worth being explicit, because each of these gets claimed on the strength of this trial:
- It does not show a benefit in people without established cardiovascular disease. Nobody like that was enrolled.
- It does not show the benefit persists after stopping. Follow-up ran while people were on the drug.
- It does not compare against other weight-loss methods. The comparator was placebo, not diet, surgery, or a different drug.
- It does not tell you anything about body composition. Weight and waist were measured; lean mass was not the subject of this trial, which is the gap our muscle loss guide deals with.
- It says nothing about tirzepatide. Different drug, different trial programme — see four names, two drugs.
What it does show, robustly, is that in adults over 45 with existing cardiovascular disease and a BMI over 27 but no diabetes, three years of weekly semaglutide produced fewer major cardiovascular events than placebo. That is a hard clinical outcome rather than a surrogate marker, and it is why this drug class is now discussed differently than it was in 2022.
It is a real finding. It is just a more specific one than the headline.
