Who loses less weight on semaglutide?

A retrospective study of 163 patients treated with semaglutide for obesity found that diabetes status, age over 60, and Hispanic ethnicity were each associated with less weight loss at 12 months^[1]. The paper, published in Endocrine Practice on 4 June 2026 by Cornet and colleagues at a US academic medical centre, adds granularity to a question the clinical trial programme left largely unanswered: who responds less well, and by how much?

What the numbers show

Of 294 patients originally prescribed semaglutide for obesity, 163 (55.4%) completed 12 months of treatment. Across those completers, mean weight loss was 8.1% of total body weight (8.1 kg, standard deviation 18.5 kg). That figure is roughly half the 14.9% mean reported in STEP-1 at 68 weeks^[2]. The gap is consistent with other real-world cohorts, where structured lifestyle support, controlled dose titration, and the selection effects built into randomised trial recruitment are absent.

On univariable analysis, three patient characteristics predicted poorer outcomes. Hispanic ethnicity was associated with less weight loss (p = 0.02; 95% CI for the coefficient: -6.82 to -0.55). A history of type 2 diabetes was associated with higher residual weight (p = 0.01; 95% CI: 0.74 to 5.76). And age over 60 was associated with less weight loss (p = 0.007; 95% CI: -14.76 to -2.41). When the authors built a multivariable model predicting failure to reach the clinically meaningful threshold of 5% total weight loss, Hispanic ethnicity remained an independent predictor (odds ratio 0.32; 95% CI: 0.14 to 0.74).

Two factors that many clinicians might expect to matter did not reach significance. Gastrointestinal side effects (p = 0.57) and provider visit frequency (p = 0.56) were both non-significant, which cuts against the assumption that better follow-up or fewer side effects automatically produce better weight outcomes.

How this fits into the wider real-world literature

The finding that people with type 2 diabetes lose less weight on GLP-1 receptor agonists is well established. Subgroup analyses from the STEP programme consistently showed smaller weight reductions in participants with type 2 diabetes compared to those without. Real-world telehealth data from Tchang et al. (2026) reported a mean 16.6% weight loss in a larger cohort of 655 patients without diabetes, with female sex as the only consistent predictor of categorical weight-loss milestones (odds ratios ranging from 2.51 to 5.90 depending on the threshold)^[3].

What the Cornet et al. study adds is an age threshold (over 60) and an ethnicity signal (Hispanic) that the randomised trial programme was not powered or designed to isolate. The STEP trials enrolled broad populations but reported subgroup analyses primarily by baseline BMI, sex, and diabetes status. Age and ethnicity were not foregrounded as effect modifiers. That makes this single-centre retrospective study a hypothesis generator rather than a confirmation, but one with plausible biological and social underpinnings. Older adults tend to have lower basal metabolic rates and more concurrent medications; the ethnicity signal likely reflects a mix of genetic, dietary, and healthcare-access factors that a 163-patient chart review cannot disentangle.

What this study cannot tell us

Several important caveats apply. The sample of 163 completers is small, drawn from a single US academic clinic, and shaped by a 45% dropout rate that the authors do not fully characterise. The retrospective design means the team observed associations, not causes. Self-reported ethnicity categories do not capture the socioeconomic, dietary, or insurance-access confounders that track closely with racial and ethnic categories in American healthcare. And the 12-month endpoint, while clinically relevant, is shorter than the 68-week window in STEP-1, making direct comparisons to the trial headline figure imprecise.

Nor does the study address the discontinuation question that looms over the entire GLP-1 weight-management field. STEP-4 showed substantial weight regain after semaglutide withdrawal, and it is not clear whether the subgroups that lose less weight initially also regain differently. That question remains open. For readers following how the GLP-1 class is evolving, the semaglutide overview and the tirzepatide comparison data on this site provide the broader trial context.

Where this lands for patients and clinicians

The practical takeaway is narrow but useful. Patients with type 2 diabetes and those over 60 should expect that their weight-loss trajectory on semaglutide may be more modest than the numbers reported from STEP-1 or STEP-3. That does not mean the drug is ineffective for these groups; an 8.1% mean weight loss still carries clinical benefit for metabolic health, joint loading, and cardiovascular risk. It means that expectations set by the 14.9% headline figure may need adjusting.

The ethnicity finding is too preliminary to shape prescribing conversations, but it flags a gap in the evidence base that larger, prospective, multi-site studies should address. Until then, the best a patient can do is discuss realistic outcome expectations with their prescriber, using real-world data alongside trial data. For more on the regulatory landscape and where semaglutide stands across different countries, see our regulation tracker. Readers may also want to review how the next-generation triple agonist retatrutide performed in TRIUMPH-1, which reported larger mean weight losses at 80 weeks.