- Design
- Mediation analysis of the SELECT cardiovascular outcomes trial
- Population
- Adults with cardiovascular disease and BMI at least 27, without diabetes, on semaglutide 2.4 mg
- Primary outcome
- Proportion of the cardiovascular benefit explained by measured risk factors
- Effect
- Joint mediation 31.4% (95% CI -30.1 to 143.6); all estimates with intervals crossing zero
In SELECT, weekly semaglutide 2.4 mg lowered major adverse cardiovascular events by 20% in adults with cardiovascular disease and a body mass index of at least 27, without diabetes. This mediation analysis asked which measured changes explained that benefit.
Semaglutide improved every candidate mediator, but none could be pinned down. The largest point estimates were for waist circumference (64%) and high-sensitivity C-reactive protein (42%), but all had very wide confidence intervals, and supplementary analyses flagged the weight and waist estimates as unreliable. The combined estimate for all mediators together was 31.4%, with an interval running from below zero to above 100%.
The practical reading is that the cardiovascular benefit is not simply weight loss, nor any single measured factor. Something beyond the standard risk markers is contributing, which matters when counselling patients that this is more than a weight drug.
- Mediation analysis of SELECT, where semaglutide cut major cardiovascular events by 20%.
- Every candidate mediator improved, but none explained the benefit with confidence.
- Point estimates were largest for waist circumference (64%) and C-reactive protein (42%).
- Combined mediation was 31.4%, with a confidence interval spanning below zero to above 100%.
- Counsel patients that the cardiovascular benefit is not reducible to weight loss alone.
Why it matters
This challenges the assumption that the heart benefit is simply a consequence of weight loss.
Don't overread it
The wide intervals crossing zero mean no single factor is confirmed as a mediator; the mechanism remains unexplained.
The statistics, in plain English
A percentage mediation whose confidence interval crosses zero and exceeds 100% is statistically uninformative: it is consistent with no mediation at all or with complete mediation. The analysis narrows nothing, which is itself the finding.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for cardiology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free