---
type: source
title: "GLP-1 receptor agonists associated with 42% lower risk of worsening mental illness in Swedish national cohort (within-individual design)"
author: "Lancet Psychiatry / Karolinska Institutet"
url: https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366(26)00014-3/fulltext
date: 2026-03-22
domain: health
secondary_domains: []
format: research-article
status: unprocessed
priority: high
tags: [GLP-1, semaglutide, mental-health, depression, anxiety, substance-use-disorder, within-individual-design, Swedish-cohort, psychiatric-safety, divergence]
intake_tier: research-task
---

## Content

Lancet Psychiatry, March 2026. Karolinska Institutet (Swedish national registry study). Study: "Association between GLP-1 receptor agonist use and worsening mental illness in people with depression and anxiety in Sweden."

**Study design:**
- n = 95,490 people with pre-existing depression and/or anxiety
- 22,480 used GLP-1 receptor agonists during follow-up
- WITHIN-INDIVIDUAL stratified Cox models — compares same person's periods ON vs. OFF GLP-1
- This design eliminates ALL time-invariant confounding (baseline psychiatric severity, comorbidities, social circumstances)
- Primary outcome: composite of psychiatric hospitalization, sick leave >14 days for psychiatric reasons, hospitalization for self-harm, death by suicide

**Key findings (semaglutide):**
- **42% lower risk of worsening mental illness** during semaglutide use vs. non-use periods (same individuals)
- Depression worsening: HR 0.56, 95% CI [0.44-0.71] → **44% reduction**
- Anxiety worsening: HR 0.62, 95% CI [0.52-0.73] → **38% reduction**
- Substance use disorder worsening: HR 0.53, 95% CI [0.35-0.80] → **47% reduction**
- Self-harm: 47% reduction

**Liraglutide:** HR 0.82 for overall mental illness worsening → **18% reduction** (weaker effect)

**Why this resolves the divergence:**
- The matched cohort study (Nature Scientific Reports, 162,253 patients) showed 195% INCREASED MDD risk in GLP-1 users
- The within-individual design demonstrates that the matched cohort finding is likely a SELECTION ARTIFACT: people prescribed GLP-1s have more baseline psychiatric comorbidity; propensity score matching cannot fully eliminate this (confounding by unobserved variables)
- Within-individual comparison strips away time-invariant confounders, showing the drug itself is associated with BETTER mental health outcomes when prescribed to people with pre-existing illness
- The FDA meta-analysis of 91 placebo-controlled RCTs (107,910 patients) also found NO increased psychiatric risk → converges with the Swedish study

**Study limitation:** Population is people with pre-existing depression/anxiety — may not generalize to patients without mental health history.

Karolinska Institutet press coverage: "Diabetes drug Ozempic linked to better mental health" — published same day.

## Agent Notes

**Why this matters:** This is the methodologically dominant evidence in the GLP-1 psychiatric safety debate. The within-individual design is the strongest quasi-experimental design available in observational epidemiology for this question — it eliminates the confounding by indication that plagues all matched cohort studies of GLP-1s. The 42-47% reductions in depression, anxiety, and SUD worsening are large effects. This should trigger a significant confidence update on GLP-1 psychiatric safety and should be documented as the study that resolves the apparent divergence.

**What surprised me:** The magnitude. A 44% reduction in depression worsening and 47% reduction in SUD worsening during semaglutide use periods — in people with pre-existing mental illness — is larger than any approved psychiatric medication for these conditions. If this effect size replicates in RCTs, GLP-1s would become first-line adjunct psychiatric medications.

**What I expected but didn't find:** Formal acknowledgment in existing KB claims that the Swedish study resolves the 195% MDD risk signal. The prior session archive (2026-05-02) documented both findings as in tension — this session clarifies that the within-individual design should dominate.

**KB connections:**
- [[GLP-1 receptor agonists are the largest therapeutic category launch in pharmaceutical history but their chronic use model makes the net cost impact inflationary through 2035]] — safety profile is better than prior signals suggested
- the mental health supply gap is widening not closing — GLP-1s addressing depression/anxiety/SUD could partially offset the supply gap through pharmacological means
- medical care explains only 10-20 percent of health outcomes — GLP-1s challenge the clean clinical/non-clinical boundary by addressing non-clinical pathways through clinical drugs
- Divergence candidate: GLP-1 psychiatric safety — matched cohort (195% MDD risk) vs. within-individual (42% reduction in worsening)

**Extraction hints:**
- Primary claim: "Semaglutide reduces worsening of depression, anxiety, and substance use disorder by 40-50% in people with pre-existing mental illness (Lancet Psychiatry, within-individual design, n=95,490)"
- Divergence document: GLP-1 psychiatric safety — within-individual vs. matched cohort design conflict
- Belief 2 complication: clinical drug achieving large effects on behavioral/non-clinical health pathways

**Context:** Published alongside Science Media Centre expert reactions. Medscape covered as "GLP-1s Tied to Lower Risk of Psychiatric Decline in Pre-existing Mental Illness." ScienceDaily: "Weight loss drug Ozempic cuts depression, anxiety, and addiction risk."

## Curator Notes (structured handoff for extractor)
PRIMARY CONNECTION: [[GLP-1 receptor agonists are the largest therapeutic category launch in pharmaceutical history but their chronic use model makes the net cost impact inflationary through 2035]]
WHY ARCHIVED: This is the methodologically strongest study on GLP-1 psychiatric safety — resolves the 195% MDD risk divergence by demonstrating confounding by indication; the within-individual finding has no precedent in the KB
EXTRACTION HINT: Write two outputs: (1) new claim on semaglutide's psychiatric protective effects (HR 0.56 for depression), (2) divergence document explaining why the matched cohort and within-individual results both exist and which to weight more heavily