Antidepressants and obesity travel together, in both directions and for several reasons, so a large share of the people buying a GLP-1 online are already taking one. The question that follows is practical rather than philosophical: does the weight loss still happen, and does anything about the combination need watching. That is a different question from whether these drugs affect mood, which the suicidality and depression evidence covers on its own terms.
The subgroup exists, and it is one post hoc analysis
The four semaglutide obesity trials that produced the headline weight figures randomized 3,683 participants, and 539 of them — about 14.6% — were taking an antidepressant at baseline. A 2024 exploratory analysis split all four trials on that variable after the fact.[1] A post hoc split is a description rather than a test; nobody randomized anyone to an antidepressant.
That distinction has teeth here, because the two groups differ by more than one prescription. People taking an antidepressant differ from people who are not by diagnosis, by how often they see a clinician, and by whatever prompted the prescription in the first place. A subgroup analysis cannot separate the medication from the condition it treats, so the result below describes what happened to a group rather than what the antidepressant did.
The weight loss did not shrink
In STEP 1, participants on an antidepressant at baseline lost a mean 15.7% of body weight on semaglutide against 0.2% on placebo. Participants not on one lost 14.7% against 2.8%. Similar patterns appeared in STEP 2, 3 and 5, and adverse events were broadly comparable between semaglutide and placebo within the antidepressant group.[1]
The interesting part is not the treated arms, which are within a point of each other and land where the main semaglutide results do. It is the placebo arms, which differ by more than the treated ones: 0.2% against 2.8%. Subtracting gives a treatment-versus-placebo gap of 15.5 percentage points with an antidepressant and 11.9 without — a difference produced mostly by the comparison group, not by the drug. Lifestyle intervention plus placebo did less for the people taking an antidepressant. That is a real observation about how hard the baseline is, and it is not evidence that semaglutide works better in this group.
The exclusion that defines the subgroup
These trials excluded anyone with severe major depressive disorder within the two years before screening, and anyone scoring 15 or higher on the PHQ-9 at screening.[1] A participant on an antidepressant who made it into the trial was therefore, by construction, someone whose treatment was working.
So the honest statement is narrow: in people with controlled depression on stable medication, the weight result holds. It says nothing about someone in an active episode, and the surrounding literature has the same shape. A 2026 systematic review of 37 studies of semaglutide in psychiatric contexts concluded that its use in mood disorders, and in depression specifically, remains controversial.[2]
The offset premise is smaller than it sounds
Behind the worry sits an assumption: that antidepressants cause weight gain substantial enough to fight the drug. A 2024 target trial emulation across eight United States health systems put numbers on it, following 183,118 patients starting one of eight first-line antidepressants and estimating weight change at six months against sertraline.[3]
The largest difference was escitalopram at 0.41 kg (95% CI, 0.31 to 0.52), then paroxetine at 0.37 kg, duloxetine at 0.34 kg, venlafaxine at 0.17 kg and citalopram at 0.12 kg. Fluoxetine was indistinguishable at −0.07 kg. Bupropion came in lower than sertraline, at −0.22 kg (95% CI, −0.33 to −0.12), and carried a 15% reduced risk of gaining at least 5% of baseline weight where escitalopram, paroxetine and duloxetine carried a 10% to 15% higher risk.[3]
Under half a kilogram separates the extremes of that list over six months. Against a drug effect measured in double-digit percentages of body weight, the choice between two antidepressants is not the variable that decides the outcome. The paper also reports six-month adherence between 28% and 41%, which caps how much any of these estimates can be attributed to the medicine at all. That is worth holding against the far larger effect of stopping the GLP-1 itself, covered in the discontinuation evidence.
Tested as a treatment, the results are mixed and small
A separate line of work asks whether a GLP-1 does anything for depression rather than despite it. A 16-week double-blind trial randomized 72 adults with major depressive disorder, cognitive impairment and overweight or obesity to adjunctive oral semaglutide 14 mg or placebo. It missed its primary endpoint: the executive function composite differed by 0.32 (95% CI, −0.92 to 1.58; p = 0.60).[4]
A preplanned secondary analysis found a difference in global cognition of 2.39 (95% CI, 0.19 to 4.60; p = 0.03) and a weight difference of −6.03 kg (95% CI, −8.76 to −3.29; p < 0.001). Depressive symptom severity did not change, and neither did the frequency of suicidal ideation.[4]A secondary report from the same trial found participants on semaglutide more willing to exert effort for reward, with sensitivity to effort reduced (β = −1.737; p = .03) and no effect on sensitivity to probability.[5]
A primary endpoint that fails while a secondary succeeds is the standard shape of a promising result that has not been demonstrated. Two secondary findings from one 72-person trial are a reason to run a larger one, not a reason to expect a mood benefit from a prescription bought for weight.
The same 2026 review that called the depression case controversial reported the opposite about adjacent conditions: it described semaglutide as effective and well tolerated across several psychiatric populations, with the strongest interest in binge eating disorder, where it noted effects on cognitive symptoms and not only on weight, and preliminary signals in alcohol and substance use disorder.[2] Those are different diagnoses with different literatures, and an appetite-driven eating disorder is not depression.
Where the psychiatric evidence is genuinely thin
The most rigorous trial in a psychiatric population was not about antidepressants at all. COaST randomized adults with schizophrenia or schizoaffective disorder on clozapine to semaglutide or placebo for 36 weeks, reaching 13.88% (SE 0.90) body weight change at week 36. Recruitment was suspended before it reached its intended 80 participants, leaving 31 randomized — 15 to semaglutide and 16 to placebo — because the investigational product was unavailable.[6]
A 2024 review of psychotropic drug-related weight gain places that in context: metformin is the best-studied countermeasure, and the evidence for GLP-1 receptor agonists in this role is described as promising and emerging rather than established, with adequately controlled studies named as the thing still needed.[7]
The interaction with a real signal is not an antidepressant
No pharmacokinetic interaction has been demonstrated between these drugs and any SSRI, SNRI or bupropion, and none of those appeared in the formal interaction program summarized in the oral medications article. Lithium is the exception in psychiatry, and it is a mood stabilizer rather than an antidepressant. A 2025 case series describes three patients on stable lithium who started semaglutide: two developed toxicity with rising levels despite unchanged renal function, and the third remained above expectation even after a preemptive dose cut.[8]
The proposed mechanisms — dehydration from reduced intake, vomiting or diarrhea, altered kidney function — describe the route by which any drug cleared renally and dosed to a blood level can drift. Three cases establish that it happens, not how often.
That contrast is the reason the absorption question sits more comfortably here than elsewhere. SSRIs, SNRIs and bupropion are not titrated against a serum concentration and do not have a narrow therapeutic index, so the effect these drugs reliably produce — a lower, later peak with total exposure intact — has no obvious way to matter for them. The medicines where a 30% shift in exposure changes an outcome are the ones with a number attached, and in psychiatry that is lithium and a handful of anticonvulsants rather than the antidepressant classes.
A caution about the reverse direction is still warranted. The interaction studies were conducted in healthy volunteers, and a person vomiting through a dose escalation is not one. A dose of any daily medicine that is brought back up is a missed dose whatever its pharmacokinetics say, and an antidepressant missed repeatedly produces discontinuation symptoms that look nothing like a drug interaction on paper.
What this leaves for an intake form
Nothing above is a reason to start, stop or change a psychiatric medication, and a change of that kind belongs to the prescriber who wrote it. What the evidence supports is that a current psychiatric medication list is information a prescriber needs before adding a weekly injection, and that a service which does not ask for one has skipped a step — one of the patterns described in the telehealth intake article. Conditions that rule the drug out entirely are listed in the contraindications article.
One market-specific point. The compounded semaglutide and tirzepatide sold by these services are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing, and no compounded product appeared in any trial above. The subgroup result described here belongs to branded semaglutide at 2.4 mg in a monitored trial, which is the standard the methodology holds every figure on this site to.