---
type: source
title: "GLP-1 Receptor Agonist Exposure Associated with 75% Lower Odds of Any Substance Use Disorder — All of Us Nested Case-Control Study"
author: "Tadesse M. Abegaz, Muktar Ahmed, Akshaya Srikanth Bhagavathula, Gabriel Frietze"
url: https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2026.1766770/full
date: 2026-03-10
domain: health
secondary_domains: []
format: article
status: unprocessed
priority: high
tags: [glp-1, substance-use-disorder, addiction, observational-study, all-of-us]
intake_tier: research-task
---

## Content

Published in *Frontiers in Psychiatry*, March 10, 2026. DOI: 10.3389/fpsyt.2026.1766770

**Study design:** Retrospective nested case-control within the All of Us Research Program. Propensity score matched 1:1 (controlling for age, sex, race/ethnicity, diabetes/obesity status, oral hypoglycemic agent use). 90-day lag period for temporal sequencing. Observation window: January 2005–February 2025. GLP-1 RAs examined: liraglutide, semaglutide, exenatide, dulaglutide (combined exposure — no individual drug differentiation).

**Sample sizes by SUD subtype:**
- AUD cohort: n=22,652
- OUD cohort: n=13,226
- NUD (nicotine) cohort: n=42,320
- CUD (cocaine) cohort: n=9,296

**Key effect sizes:**
- Any SUD combined: **OR = 0.25** (95% CI 0.22–0.30) — 75% lower odds
- AUD: **OR = 0.26** (95% CI 0.20–0.34) — 74% lower odds
- OUD: **OR = 0.31** (95% CI 0.23–0.42) — 69% lower odds
- NUD: **OR = 0.32** (95% CI 0.27–0.39) — 68% lower odds
- CUD: **OR = 0.25** (95% CI 0.16–0.40) — 75% lower odds

**Key limitations:** Observational — reverse causality possible despite 90-day lag. Combined GLP-1 exposure (individual drug effects not differentiated). Unmeasured confounding from psychiatric comorbidity, social support, healthcare-seeking behavior. No causal mechanism established.

**Context in converging evidence:**
This is the third independent evidence stream for GLP-1 and SUD reduction:
1. This study (All of Us, observational): OR=0.25, 75% lower odds — strongest effect size, weakest design
2. Lancet Psychiatry Swedish cohort (within-individual, n=95,490): 47% SUD worsening reduction — strongest design for causal inference
3. JAMA Psychiatry RCT (2025): 41% reduction in heavy drinking days, NNT 4.3 — gold standard design, AUD + obesity

## Agent Notes
**Why this matters:** The convergence of three independent designs — with consistent direction despite different populations, methods, and outcome definitions — is the strongest pattern in the GLP-1 psychiatric evidence base. Even with observational limitations, OR=0.25 across four distinct substance categories (alcohol, opioid, nicotine, cocaine) is hard to explain as pure selection bias.

**What surprised me:** The cocaine use disorder effect size (OR=0.25) is as large as the alcohol effect. There is NO behavioral intervention that produces 75% reduction in cocaine use disorder odds. If this holds up to causal scrutiny, it would represent the largest treatment effect for CUD in the literature.

**What I expected but didn't find:** Individual GLP-1 drug differentiation (semaglutide vs. liraglutide vs. dulaglutide). The lack of drug-level analysis is a significant limitation given the mechanistic differences between GLP-1 agonists.

**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]]
- Connects to Session 34-35 findings on GLP-1 + AUD (NNT 4.3, JAMA Psychiatry 2025)
- Connects to Lancet Psychiatry Swedish cohort (Sessions 37-38) — third independent evidence stream

**Extraction hints:**
1. Claim about GLP-1 reducing SUD odds across all substance categories — the cross-category breadth is the most distinctive feature
2. The convergence across three designs is itself a meta-claim worth writing
3. The cocaine use disorder effect size may be worth a specific claim given the absence of any comparable intervention

## Curator Notes (structured handoff for extractor)
PRIMARY CONNECTION: [[GLP-1 receptor agonists are the largest therapeutic category launch in pharmaceutical history]]
WHY ARCHIVED: Provides the largest observational evidence base for GLP-1 and SUD reduction — converges with within-individual Swedish study and JAMA Psychiatry RCT to form a three-design convergence pattern
EXTRACTION HINT: Lead with the convergence pattern across three designs, not just this study — the extractor should write a claim that synthesizes all three evidence streams, then evaluate confidence based on their convergence