Baseline features barely predict who loses most weight on semaglutide

Design

Prespecified secondary analysis of a randomised, double-blind trial

Population

7,594 adults with obesity and atherosclerotic disease, without diabetes, on semaglutide ≥1 year

Primary outcome

Peak percentage body weight loss

Effect

Median 12.5% (IQR 7.9–17.9) at median 19 months; baseline model explained <10% of variance

This prespecified analysis of SELECT took the 7,594 participants who stayed on semaglutide for at least a year — 86.3% of those randomised to it — and asked whether anything measurable at baseline predicted their peak weight loss. These were adults without diabetes, mean age 62, 72% men, all with established atherosclerotic disease.

Median peak weight loss was 12.5% (IQR 7.9–17.9), and it took a median of 19 months to reach it. The full model of baseline demographics, history, laboratory values and measurements explained less than 10% of the variation, and female sex accounted for most of that. Women reached a median of 16.6% against 12.4% for men. The largest visible difference was dose: those whose maximum dose was 0.24 mg lost a median 6.0%, against 13.3% for those who reached 2.4 mg.

In practice this removes a common excuse for early pessimism. You cannot tell a patient at the first visit whether they will be a good or poor responder, and the plateau comes late. The modifiable levers are reaching and holding the target dose, and giving the drug long enough.

• Do not predict a patient's semaglutide response from their age, BMI or labs at the first visit — baseline features explained under 10% of it.
• Plan for peak weight loss at around 18–24 months, not at the 3-month review.
• Titrate to the full 2.4 mg obesity dose where tolerated; patients left at low doses lost about half as much.
• Expect women to lose more on average than men, but individual spread is wide in both.
• These data come from adults without diabetes; people with type 2 diabetes typically lose less on the same drug.

Why it matters

It shifts the conversation from predicting responders to persisting with dose and time before judging failure.

Don't overread it

The dose comparison is observational within the treated arm — lower-dose patients may differ in tolerance and adherence.

The statistics, in plain English

"Explained less than 10% of the variability" means that if you knew every baseline characteristic in the model, you would still be almost entirely guessing an individual's weight loss. The interquartile range (7.9–17.9%) shows the middle half of patients, so a quarter lost less than 8% and a quarter more than 18%. The dose comparison is not randomised — people who stayed on low doses may have done so because of side effects or poor adherence — so the gap between 6% and 13% overstates what dose alone does.

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.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.