Yes, by about 10.1 kg, and roughly half of it came back. A 24-week open-label study gave nurse-administered semaglutide to people with schizophrenia taking clozapine or olanzapine [1]. Weight fell 9.8% on an intention-to-treat basis (95% CI −12.7% to −6.8%). A year after the drug stopped, participants were 5.1% below baseline. Twenty-six people enrolled and fifteen finished the trial, with no control group.
Clozapine and olanzapine cause substantial weight gain. For many people they are also the antipsychotics that work, so they are not substitutable. That leaves a population carrying drug-induced obesity with no way to stop the cause.
What the intervention produced
Weight fell 9.8% on an intention-to-treat basis, around 10.1 kg. Waist circumference fell 7.3%. HbA1c fell too but did not reach significance (P = 0.055).
The dose was 1.0 mg weekly. That is a diabetes dose, not the 2.4 mg used for weight management. These figures are therefore not comparable to obesity-trial numbers, a distinction that changes the ranking of these drugs. The ordering is set out in the network meta-analysis of nineteen drugs.
How small the study is
Twenty-six people enrolled. Seventeen completed the intervention and fifteen completed the trial. There was no control group and no blinding.
In a population where medication adherence is the central clinical difficulty, a 65% completion rate is itself a finding. It means the weight figures describe those who stayed.
One result points in an unexpected direction. Gut microbial alpha diversity decreased as time on semaglutide increased, with one species enriched. Lower diversity is usually characterized as unfavorable in metabolic research. Nobody measured a health outcome from it, so it is an observation rather than a harm. It belongs beside the other things these drugs turn out to do that nobody designed them for, as in the inflammation substudy.
What happens to mortality in this population
People with schizophrenia, bipolar disorder and major depression die substantially earlier than everyone else, mostly of cardiovascular disease. A target trial emulation matched 764,115 pairs starting a GLP-1 drug or an SGLT2 inhibitor. Of those pairs, 195,184 had serious mental illness [2].
Four-year mortality in that cohort was 4.91% against 6.45%. The hazard ratio was 0.76 and the absolute difference 1.54 percentage points. The comparator is not placebo: SGLT2 inhibitors reduce mortality themselves, so this is one effective drug measured against another. The same paper's one-year figure is 1.46% against 2.84%. That is too large to believe, and the reasons are in the four-year mortality gap.
Whether the psychiatric course changes
A nationwide cohort covered 14,694 people with bipolar disorder and diabetes or obesity. It compared each person against their own periods on and off these drugs [3]. Semaglutide tracked a 21% lower risk of psychiatric hospitalization (adjusted HR 0.79, 95% CI 0.69 to 0.91). Liraglutide and dulaglutide showed no such association.
One outcome did not move at all. Sick leave for psychiatric reasons showed no association with any of the drugs studied. Hospitalization depends on admission thresholds and bed availability; sick leave tracks whether someone can work. The full reading is in the bipolar cohort.
One randomized psychiatric result exists. A trial in major depressive disorder found semaglutide reduced the felt cost of effort, measured on an effort-based decision-making task [4]. It was a secondary finding, and it is described in the effort and reward trial.
What this establishes
Feasibility, in a group usually excluded from obesity trials altogether. Nurse-administered treatment in a public mental health setting produced real weight loss in people taking the most metabolically damaging antipsychotics.
Whether it should be continued indefinitely rather than for 24 weeks is the question the 76-week figure raises and cannot answer. The same question runs through what comes back when you stop. No seller on this roster prescribes for schizophrenia, and an intake form is not where a clozapine history belongs.