News · Heart & Metabolic
Semaglutide led the field for antipsychotic weight gain
A JAMA Psychiatry network meta-analysis of 95 trials ranked 39 drugs for reversing antipsychotic weight gain, and semaglutide came out well ahead at nearly 11 kg.
Based on a peer-reviewed network meta-analysis in JAMA Psychiatry
- Researchers in Toronto pooled 95 randomized trials testing 39 different drugs, covering 5,898 participants, in JAMA Psychiatry.
- All participants had a schizophrenia spectrum disorder and were taking antipsychotics, which commonly cause substantial weight gain.
- Semaglutide produced the largest reduction, a mean 10.98 kg below placebo (95% CI, -13.33 to -8.62), from three trials, at moderate certainty.
- Liraglutide followed at 5.43 kg, then topiramate at 3.95 kg, metformin at 3.86 kg and exenatide at 2.97 kg, all at moderate certainty.
- Only semaglutide and metformin reached clinically meaningful weight change of 5% or more.
- Metformin rests on much more evidence, 16 trials against semaglutide's 3.
- Other agents showed effects at very low certainty, meaning the estimates could easily change.
- No major concerns emerged for gastrointestinal side effects or dropout between interventions.
Antipsychotics work, and they make people gain a great deal of weight. That trade-off drives much of the cardiometabolic illness that shortens life in schizophrenia. Clinicians have long lacked a clear ranking of what to do about it. An analysis in JAMA Psychiatry, published by the American Medical Association, supplies one.
A total of 95 studies examining 39 individual pharmacological interventions were included. Semaglutide came out well ahead of everything else.
The gap this fills
The problem is not new. Significant weight gain is a concerning adverse effect of antipsychotic medications experienced by patients with schizophrenia spectrum disorders. Its high prevalence and significant contribution to cardiometabolic morbidity in this population warrant better consensus on the management.
The obstacle has been fragmentation: dozens of small trials, few compared directly. A network meta-analysis is built for that situation. It uses shared placebo comparisons to place drugs on one scale.
What went in
The team searched seven databases and trial registries up to December 2025, including randomized clinical trials examining any pharmacological intervention for weight reduction in antipsychotic-treated patients. Trial length was not restricted.
A total of 95 studies examining 39 individual pharmacological interventions were included, pooling 5898 participants. Certainty was graded with a formal tool rather than left to impression.
The ranking
Five agents separated from the field at moderate certainty. Semaglutide, liraglutide, topiramate, metformin, and exenatide were associated with the most significant reductions in body weight compared to placebo, at 10.98, 5.43, 3.95, 3.86 and 2.97 kg respectively.
The gap between first and second is unusually wide for this kind of analysis: semaglutide roughly doubled the next best result.
But raw kilograms are not the only test. Clinically meaningful weight change of 5% or greater was observed with semaglutide and metformin alone.
Other interventions including ramelteon, nizatidine, and aripiprazole were also found to be associated with weight-reducing effects but with very low certainty of evidence. That is the paper’s polite way of saying do not rely on those numbers.
Reading the ranking properly
The obvious headline is semaglutide. The obvious caution is that it rests on three trials, while metformin rests on sixteen. Both cleared the clinically meaningful threshold. One is cheap, familiar and heavily evidenced; the other is more powerful on thinner ground.
Network meta-analysis also carries its own fragility. Comparing drugs that were never tested against each other assumes the trials were similar enough to link. With 95 studies spanning decades and 39 agents, that assumption strains.
Tolerability at least looked unremarkable: there were no major concerns with gastrointestinal adverse effects or leaving the study early between interventions.
The authors position the work as guidance, not instruction. These agents were associated with the greatest reductions in body weight and were supported by the highest-certainty evidence, providing guidance for clinicians managing antipsychotic-associated weight gain.
For patients, the message may be simpler than the ranking. This side effect has treatments, several of them. It does not have to be accepted as the price of staying well.
People also ask
Why is antipsychotic weight gain such a problem?
It is common, substantial and consequential. Significant weight gain is a well-recognized adverse effect of antipsychotic medications, and it contributes heavily to the cardiometabolic illness that shortens life in this group. It is also a leading reason people stop taking medication that is otherwise controlling their symptoms, which makes it a psychiatric problem as well as a physical one.
What is a network meta-analysis?
A method for ranking treatments that have rarely been tested head to head. If drug A beat placebo in one trial and drug B beat placebo in another, a network analysis uses those shared comparisons to estimate how A and B compare with each other. It is more informative than a simple pooled average, and more fragile: the indirect comparisons rest on assumptions that the trials were similar enough to link.
Semaglutide won, so why mention metformin?
Because evidence volume matters alongside effect size. Semaglutide's 10.98 kg came from 3 trials; metformin's 3.86 kg came from 16. Both were rated moderate certainty and both reached the clinically meaningful 5% threshold. Metformin is also far cheaper and long familiar in this setting. A larger effect on thinner evidence is not automatically the better bet.
What does 'very low certainty' mean for the other drugs?
That the estimate should be treated as close to uninformative. Ramelteon, nizatidine and aripiprazole all showed weight-reducing effects, but at very low certainty in the authors' assessment, meaning future trials could easily overturn them. Reporting the ranking without the certainty rating is how these rankings get misread.
Should someone on antipsychotics ask for one of these?
It is a reasonable conversation to have, and it is a conversation, not a self-prescription. This is general information rather than advice. Antipsychotic weight gain is treatable and under-treated, so raising it with a psychiatrist is worthwhile. Any addition has to be weighed against interactions, existing conditions and whether the antipsychotic itself could be switched instead.