Antipsychotic-induced weight gain drives most of the cardiometabolic mortality gap in schizophrenia, and until now the choice of what to do about it has rested on scattered small trials. This network meta-analysis assembled 95 randomised trials covering 39 pharmacological interventions in 5,898 antipsychotic-treated patients with schizophrenia spectrum disorders, with certainty assessed using CINeMA.
Five agents stood out, all at moderate certainty. Semaglutide reduced weight by 10.98 kg against placebo (95% CI -13.33 to -8.62, from 3 trials), liraglutide by 5.43 kg (-8.54 to -2.33, 2 trials), topiramate by 3.95 kg (-5.89 to -2.02, 5 trials), metformin by 3.86 kg (-5.02 to -2.70, 16 trials) and exenatide by 2.97 kg (-5.83 to -0.11, 3 trials). Only semaglutide and metformin achieved a clinically meaningful weight change of 5% or more. Several agents also improved other metabolic outcomes, and there were no major differences in gastrointestinal adverse effects or dropout.
Ramelteon, nizatidine and aripiprazole also showed weight-reducing effects, but at very low certainty, and should not be chosen on this evidence.
The practical ranking is metformin first and semaglutide where affordable. Metformin rests on 16 trials — by far the largest evidence base here — costs almost nothing, and delivers a clinically meaningful reduction. Semaglutide delivers nearly three times as much weight loss but on three trials and at a price that puts it out of reach for most patients with severe mental illness in Indian practice, where this population is among the least able to pay.
And none of this replaces choosing a less obesogenic antipsychotic in the first place where the psychiatric indication allows.
- Start with metformin: 16 trials, moderate certainty, clinically meaningful weight loss, and negligible cost
- Use semaglutide where affordable — nearly 11 kg against placebo, the largest effect in the network
- Do not choose ramelteon, nizatidine or aripiprazole for this purpose; the evidence is very low certainty
- Consider switching to a less obesogenic antipsychotic before adding a second drug, where the psychiatric picture allows
- Measure weight, waist, glucose and lipids at baseline and at intervals — the intervention only follows the measurement
The statistics, in plain English
Read the k values alongside the effect sizes. Semaglutide's -10.98 kg rests on 3 trials; metformin's -3.86 kg rests on 16. A large effect from few trials is more likely to shrink with further data than a modest effect from many, which is why the two should not be compared on point estimate alone. Exenatide's interval, -5.83 to -0.11, almost touches zero and should be read as weak. Network meta-analysis compares treatments that were never tested head to head by linking them through shared comparators, which introduces assumptions the CINeMA certainty ratings are trying to quantify — the moderate ratings mean the estimates are probably close to the truth but could change.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free