Design: Prespecified analysis of the SELECT cardiovascular outcomes trial
Population: 7,594 adults with obesity and atherosclerotic disease, without diabetes, on semaglutide for at least a year
Primary outcome: Baseline predictors of peak percentage body-weight loss
Effect: Median peak loss 12.5% at about 19 months; baseline features explained under 10% of the variability

This prespecified analysis of SELECT asked whether routinely available baseline features predict how much weight a person loses on semaglutide. It looked at the 7,594 participants who stayed on the drug for at least a year, from a trial of adults with obesity and atherosclerotic disease but without diabetes.

Median peak weight loss was 12.5%, reached after a median of about 19 months. A model using all the candidate baseline predictors explained under 10% of the variation between individuals, and female sex accounted for most of even that small share. Dose mattered: those who reached 2.4 mg lost a median 13.3% against 6.0% for those capped at 0.24 mg, and women on the top dose lost the most.

The honest conclusion is that you cannot tell at the outset who will respond best. Set expectations on the average and keep titrating to the target dose rather than judging a patient on early numbers.

Prespecified analysis of 7,594 SELECT participants on semaglutide for at least a year.
Median peak weight loss was 12.5%, reached after a median of about 19 months.
Baseline characteristics together explained under 10% of the variation in weight loss.
Women on the 2.4 mg dose lost the most (median 16.6%); men lost less (median 12.4%).
Participants had obesity and cardiovascular disease but not diabetes.

Why it matters

Set expectations on the average, not on a prediction; you cannot tell at the start who will lose most.

The statistics, in plain English

A model explaining under 10% of the variation means baseline features are almost useless for predicting one person's result. The wide spread around the median (interquartile range 7.9-17.9% overall) shows how differently individuals respond.