SELECT — trial digest

SELECT (N Engl J Med 2023): Does semaglutide reduce major adverse cardiovascular events in people without diabetes?

Citation and design

This digest summarises the trial and the criticisms our journal club considered material. It is not a substitute for the paper, and where the two differ the paper is right.

The question it set out to answer

Does semaglutide reduce major adverse cardiovascular events in people without diabetes?

Stating the question before the result is a house convention. A great many arguments about this trial are actually arguments about what it was for.

What it found

The first cardiovascular outcome trial in this class in a population without diabetes, and the most consequential result in the field for how the drugs are positioned.

The primary composite outcome was reduced. The relative reduction is the figure that circulates; the absolute difference between arms is the figure that determines how many people benefit.

The result decoupled the cardiovascular argument from glycaemic control, which is why it mattered beyond its own numbers.

Criticisms the journal club considered material

1. The population had established cardiovascular disease, so absolute benefit in a lower-risk population would be smaller even at the same relative effect.
2. Weight reduction, blood pressure, lipids and inflammation all changed; the trial cannot apportion the benefit between them.
3. Reporting emphasised the relative reduction. Restating it as an absolute difference and a number needed to treat is the single most useful correction available in this discussion.

These are criticisms of the design and of how the result is used, not of the investigators. Sessions that lose that distinction get moderated.

How it is misused

The two recurring misuses of this trial in general discussion are quoting the relative effect without the absolute one, and extending the result to a population the trial did not enrol. Both are easy to correct and both keep happening.

Where a number appears in secondary coverage without an estimand attached, assume it is whichever version is larger and check the paper.

Maintenance

Digests are reviewed when new analyses of the same trial are published, and at the review date shown above regardless. If a secondary publication changes the picture, raise it in journal club rather than editing the digest directly, so the change is discussed before it lands.