What predicts semaglutide weight loss?

A new analysis of the SELECT cardiovascular trial, published in Diabetes Care on September 24, 2026, asked a narrow but practical question: can a doctor predict how much weight a given patient will lose on semaglutide before they even start treatment^[1]? Lead author Ildiko Lingvay and colleagues looked at 7,594 SELECT participants who stayed on semaglutide for at least a year, and the baseline characteristics a clinic can measure at the first visit, age, weight, medical history, lab values, explained less than 10% of how much weight any one person eventually lost.

Where the SELECT data comes from

SELECT (Semaglutide Effects on Cardiovascular Outcomes in People with Overweight or Obesity) enrolled 17,604 adults aged 45 and older with a body mass index of 27 or higher, established cardiovascular disease, and no history of diabetes^[2]. Novo Nordisk sponsored the trial and registered it as NCT03574597. Participants took weekly subcutaneous semaglutide at a target dose of 2.4 mg or a matching placebo. The trial's primary result, published by Lincoff and colleagues in the New England Journal of Medicine in November 2023, found a 20% reduction in cardiovascular death, non-fatal heart attack, or non-fatal stroke over a mean follow-up of 39.8 months (event rate 6.5% on semaglutide versus 8.0% on placebo, hazard ratio 0.80). That result later supported the FDA's March 2024 decision to add a cardiovascular-risk-reduction indication to the Wegovy label^[3], and the EMA authorises the same use for Wegovy across the EU and EEA^[4]. The new predictors paper is a prespecified secondary analysis of that same trial population, not a separate study.

The numbers on peak weight loss

Among the 7,594 participants randomized to semaglutide who continued treatment for at least a year (86.3% of everyone assigned to the drug), median peak weight loss was 12.5% of body weight, reached after a median of 19 months on treatment. The researchers then built a statistical model using every baseline demographic, medical-history, laboratory, and clinical measurement they had on file as candidate predictors of that peak loss. The model came up mostly empty: it accounted for less than a tenth of the variation between patients.

Two factors did show a measurable association. Sex was the strongest one: women reached a median peak loss of 16.6% of body weight, against 12.4% for men, and sex alone accounted for roughly 8 percentage points of the model's already-small predictive power. Dose mattered too. Participants whose maximum tolerated dose topped out at the trial's starting dose of 0.24 mg lost a median of 6.0%, compared with 13.3% among those who reached the trial's target dose of 2.4 mg. Neither finding is surprising by itself. Higher drug exposure producing more weight loss, and women responding more strongly to GLP-1 receptor agonists, have each shown up in earlier semaglutide trials. What stands out is how little else the model could explain.

What did not predict the outcome

Age, baseline body weight, cardiovascular risk factors, and the other laboratory and clinical measurements the authors tested were not meaningfully associated with peak weight loss once sex and dose were accounted for. The authors' own conclusion is direct: the amount of weight a person will eventually lose on semaglutide cannot be predicted from the characteristics available in a clinic visit. They attribute the unexplained variation to biological, behavioral, or genetic differences that current testing does not capture.

What we still don't know

This is a secondary analysis of one trial population, and that population was specific: mean age 62, 72% men, everyone had established atherosclerotic disease and obesity, and nobody had diabetes. It is a different group from the SUSTAIN and STEP programme trials that established semaglutide's diabetes and weight-loss indications, so these particular numbers should not be read onto a younger patient, someone without heart disease, or someone using Wegovy for weight loss alone. The analysis also cannot say why sex and dose matter, only that they correlate with the outcome in this dataset. And because it only reports on patients who stayed on treatment for a year or more, it says nothing about the roughly 14% who stopped earlier.

Why this matters

For a reader considering semaglutide, or already on it, the honest takeaway is that individual response varies widely and a clinic cannot forecast it from a chart review. Reaching the trial's target dose was associated with more weight loss, which is one more reason the slow up-titration schedule on the Wegovy label exists: it is designed to get patients to that dose safely rather than skip ahead. Beyond that, a single number quoted from a trial average, whether it is SELECT's median or STEP-1's mean, describes the middle of a wide range, not a personal forecast. That conversation, and any decision about starting or adjusting treatment, belongs with a prescribing clinician who knows the patient's full history.