The weight evidence exists and it is measured across eight randomized trials in 664 antipsychotic-treated patients. Adding a GLP-1 reduced body weight by 6.74 kg against control, with an interval of 10.05 to 3.43 [1]. The psychiatric safety question is the one that is still open, and the abstract answers it in a sentence its own results do not support.
Antipsychotics cause weight gain, and that weight gain is one of the reasons people with schizophrenia and bipolar disorder die earlier than everybody else. A drug that reverses some of it is genuinely worth pooling the evidence on. This meta-analysis does that competently, and then the conclusion says one thing more than the results support — which matters here because this is the population that already receives the least monitoring.
What was measured
Eight randomized trials were pooled, covering 664 patients taking antipsychotics. [1] Against control, adding a GLP-1 drug reduced body weight by 6.74 kg, with an interval of 10.05 to 3.43, and reduced body mass index by 2.37 kg/m². Waist circumference fell 4.27 cm, HbA1c fell 0.61 percentage points, and fasting glucose fell 6.82.
Those are the outcomes. All five are cardiometabolic, and all five moved in the direction anyone would hope. Eight small trials pooled together is a thinner base than a single large one, the way five studies in 182 people are thinner than their combined headline suggests.
The side effect that did show up
Nausea, vomiting and constipation all rose significantly, each at p below 0.01. That is the familiar pattern and nobody should be surprised by it.
Overall adverse events came out at p = 0.77, serious ones at p = 0.09, and stopping because of them at p = 0.20. Read those as what they are: 664 patients is far too few to detect a difference in rare events. A non-significant result there tells you the trials were small, not that the drug was clean. It is the same trap as reading a failure to find a difference as proof of equivalence.
The psychiatric evidence that does exist is molecule-specific
A Swedish national cohort of 95,490 people with depression or an anxiety disorder analyzed four GLP-1 drugs one at a time [2]. Against non-use, semaglutide carried an adjusted hazard ratio of 0.58 (95% CI 0.51–0.65) for worsening mental illness and liraglutide 0.82 (0.76–0.89), while exenatide and dulaglutide both sat at 1.01.
That population is not this one — those people had depression or anxiety and were on antidiabetic medication, not antipsychotics — and the lesson still carries. When two members of a class show nothing and two show a large association, a class-level claim of psychiatric safety is close to meaningless, and tirzepatide was in neither analysis.
The semaglutide number
Semaglutide showed the largest reductions, 11.06 kg of weight and 3.60 kg/m² of BMI, with the between-subgroup difference reaching significance.
That comparison was assembled by sorting eight trials into groups by which drug they used, not by randomizing anybody between drugs. Trials using different molecules also differ in their patients, their durations and their doses, so the molecule gets credit for all of it. That is exactly why indirect drug comparisons keep producing winners that direct ones do not confirm.
What monitoring this assumes
Everything above describes trial conditions, where somebody is watching. Across 49 studies and 5.5 million people, those carrying a diagnosed mental disorder were less likely to receive recommended diabetes monitoring, at an odds ratio of 0.81 (95% CI 0.70–0.94) [3]. HbA1c testing, retinal screening, cholesterol measurement and foot examination all pointed the same way.
A cash-pay prescription arrives with less of that infrastructure rather than more, which is the part of this decision a price page cannot help with. What follow-up a seller actually provides is worth asking before ordering, alongside what patients report about how much support varies.
If you are on an antipsychotic
Talk to whoever prescribes it before adding anything, because interactions and monitoring here are genuinely a specialist matter and nothing on a website substitutes for that.
What you can take from this is that the weight evidence in this population is real and reasonably consistent, and that the gastrointestinal cost is real too. The psychiatric safety question is still open, despite a conclusion that sounds like it is closed. Larger and longer trials are what the authors ask for, and it is the right ask — the same gap that leaves plenty of questions about these drugs unanswered.
On cost, nothing here changes the arithmetic. This is an add-on to medication somebody is already paying for, and the value question gets no easier when a second prescription joins the first.