GLP-1s and Psychiatric Medications

This is the group with the largest unmet need and the least attention, and I want to start with the number that frames it.

People with serious mental illness die 10–20 years earlier than the general population, and the largest contributor is not suicide. It is cardiovascular and metabolic disease. Some of that is the illness, some is access to care, and a meaningful share is the pharmacology of the treatments that keep people well.

The weight gain is a drug effect

This should not need saying and it does. Several of the most effective psychiatric medications cause substantial weight gain through appetite and metabolic effects — not through people becoming careless.

A large comparative analysis of 18 antipsychotics found wide differences in metabolic impact across the class:

Among antidepressants, mirtazapine and paroxetine are more associated with gain than most SSRIs. Valproate and lithium contribute. Some anticonvulsants used as mood stabilisers do too.

Telling a person on clozapine to eat less has a well-documented record of failure, and it moves the responsibility onto the patient for a pharmacological effect they did not choose. The person taking it usually knows this better than the clinician saying it.

Where a GLP-1 fits

Directly, and with an important boundary.

Metformin has been the conventional first-line addition for antipsychotic-associated weight gain, with a reasonable evidence base across trials of 12–24 weeks, though the average reduction is modest. GLP-1 receptor agonists are increasingly used in this setting, and the mechanism — appetite suppression against a drug-driven appetite increase — addresses the actual problem rather than working around it.

The boundary: do not stop or reduce the psychiatric medication to manage weight. Relapse of serious mental illness is a much larger threat to your life than the weight is, and switching within the class is a decision for your psychiatrist, made with the knowledge of what has and has not worked for you before. Sometimes a switch is possible. Frequently it is not, and treating the weight directly is the better answer.

Interactions, honestly assessed

These drugs are not metabolised through the cytochrome pathways where most psychiatric interactions live, so classical interactions are limited. What matters is delayed gastric emptying, which alters the rate of oral absorption.

That is clinically irrelevant for most psychiatric medication and worth attention for a few:

• Lithium — narrow therapeutic index, monitored in a range where a shift of 0.2 mmol/L matters. Levels also shift with fluid and sodium status, which both change on this drug, and shift further during any illness with vomiting or diarrhea. Monitoring through titration and during sick days is not optional; the general framework is in sick day rules.
• Anticonvulsant mood stabilisers with narrow ranges — worth flagging to whoever monitors levels.
• Anything you were told to take with food, where “with food” now means considerably less food.

The broader list is in GLP-1s and your other medications. Take your full list to a pharmacist; it takes five minutes and almost nobody does it.

Mood, and how to think about a change

Regulatory reviews by the FDA and the EMA did not find evidence of a causal link between these drugs and suicidal ideation, and large observational analyses have not shown increased risk.

That is reassurance about causation, not a reason to ignore symptoms. Depression is common, this population starts with an elevated baseline rate, and a year of rapid physical change is destabilising for anyone. The flatness some people describe around months 3–5 is discussed in food noise, mood, and the quiet months.

The practical rule: whoever manages your psychiatric treatment should know you have started. Not as a formality — so that if something changes, it gets assessed on its merits rather than attributed to whichever drug is newest.

Eating disorders

Binge eating disorder is common, frequently undiagnosed, and often improves with these medications — the reduction in food preoccupation many people describe is directly relevant.

Restrictive eating disorders are the opposite case. These drugs produce exactly what those disorders reward: very low intake, rapid loss, indifference to food. A history of anorexia or a restrictive pattern needs specialist input before starting, not a screening checkbox.

Practical points that get missed

Intake can fall too far. In someone whose appetite was pharmacologically elevated and is now pharmacologically suppressed, the swing can be large. Protein at 1.2–1.6 g/kg a day matters here as much as anywhere, and 2 sessions a week of resistance training with it — what to actually eat.

Sedation plus reduced intake is a fatigue combination. Check total intake before concluding anything — under about 1,000 calories a day will do this on its own. Check that low energy is not simply under-eating before concluding anything about the psychiatric treatment.

Metabolic monitoring should continue and probably improve. Glucose, lipids and blood pressure are already meant to be monitored on antipsychotics, typically at 3 months and then annually and frequently are not — reading your own labs covers what to ask for.

Antidiabetic and antihypertensive doses will need reviewing every 3–6 months as weight falls, which is the same deprescribing conversation as everyone else’s.

What I would want in place

• The psychiatrist and the prescribing clinician both know about each other
• Lithium or anticonvulsant levels monitored through titration and after any dehydrating illness
• A written protein target, because intake can drop a long way here
• No change to psychiatric medication made in pursuit of weight without the person who manages it
• An agreed plan for who to contact if mood changes, so that it is assessed rather than debated

The goal is not to trade psychiatric stability for a smaller body. It is to stop making people choose between the treatment that keeps them well and the metabolic consequences of taking it — which, for a long time, is exactly the choice we offered.