# What a GLP-1 drug comparison really shows

> A network meta-analysis ranked newer incretin drugs higher for average weight loss, but much of the ranking is indirect and long-term safety remains uncertain.

_Source: Peer-reviewed BMJ Medicine network meta-analysis, verified against the open full text, PubMed/Europe PMC metadata, and the US FDA tirzepatide label · 2026-10-15 · 8 min read · Verified against primary sources_

Canonical: https://iyu.app/e/glp1-weight-loss-network-meta-analysis

## The 60-second version

Across 58 randomized trials, newer incretin treatments produced the largest average weight-loss estimates, but much of the comparison was indirect.

**Key points**

- The analysis included 24,214 adults with overweight or obesity who did not have diabetes.
- Versus placebo, estimated average weight changes were -22.10% for retatrutide, -19.28% for tirzepatide and -17.32% for CagriSema.
- Several confidence intervals overlapped, and some newer agents were represented by only one or a few trials.
- Adverse-event discontinuation comparisons were less certain and cannot substitute for drug-specific safety data.
- A statistical ranking can include investigational drugs and does not determine the best treatment for an individual.

**Verdict.** The study maps a strong efficacy signal for newer incretin drugs, but approval status, long-term safety and patient context still govern real treatment decisions.

## Full explainer

A network meta-analysis of **58 randomized trials and 24,214 adults without diabetes** found the largest average weight-loss estimates with newer incretin drugs. The result is a comparison of trial networks, not proof that one drug is best for every patient.

> **⚑ Caveat:** The headline percentages are estimates versus placebo, not direct results from one trial comparing every drug. Several confidence intervals overlapped, and some newer agents were supported by only one or a few trials.


### Review — What the researchers combined

The [BMJ Medicine study](https://doi.org/10.1136/bmjmed-2026-003026) searched Embase, PubMed and Web of Science from January 2000 through 6 March 2026. It included randomized trials lasting at least 12 weeks in adults with overweight or obesity, while excluding trials with diabetes or unclear diabetes status.

The authors modelled doses separately and compared GLP-1 receptor agonists with related dual- and triple-receptor treatments. A network model can estimate differences through a shared comparator such as placebo even when two drugs have not been tested directly against each other.


### Results — The largest estimates came from newer treatments

- **-22.10%** — retatrutide versus placebo; 95% CI -25.60 to -18.60
- **-19.28%** — tirzepatide versus placebo; 95% CI -20.39 to -18.16
- **-17.32%** — CagriSema versus placebo; 95% CI -19.32 to -15.32

These are average percentage changes in body weight estimated across the network. Conventional GLP-1-only agonists generally showed smaller effects, while waist circumference and lipid outcomes followed a broadly similar pattern.

> A network ranking is a map of average trial evidence, not a prescription list.


### Method — Indirect comparison adds reach and uncertainty

- **Direct evidence:** Two treatments are randomized against each other within the same trial.
- **Indirect evidence:** Two treatments are connected through a shared comparator, often placebo.
- **Network estimate:** Direct and indirect evidence are combined if trial populations and designs are sufficiently comparable.
- **What can shift a rank:** Dose, duration, participant mix, titration, adherence and sparse evidence can all change an estimate.

The authors found overlapping confidence intervals in several comparisons and warned that newer drugs sometimes rested on only one or a few studies. Long-duration estimates often came from separate trials rather than follow-up of the same participants.


### Safety — Discontinuation is not a complete safety profile

The model suggested higher rates of stopping because of adverse events with danuglipron and retatrutide, and lower rates with mazdutide. But this outcome showed network inconsistency, and some comparisons were based on low-certainty evidence.

A discontinuation measure combines very different experiences and cannot separate persistent nausea from rare serious organ-specific events. Most included trials were not powered to detect rare harms, and long-term safety evidence for newer agents was limited.


### Status — A high rank does not mean a drug is approved

The network includes investigational and marketed treatments. The current [US FDA label](https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22ZEPBOUND%22&limit=1) confirms tirzepatide for chronic weight management in eligible adults alongside diet and physical activity; it also lists common gastrointestinal reactions and important warnings. Retatrutide had no record in the FDA approved-drug dataset we checked.

> **i** Approval, indication and availability vary by country and can change. This article explains evidence and is not a recommendation to start, stop or switch treatment.


### Limits — What the analysis cannot decide

- **1. No universal winner:** average trial effects do not identify the best option for an individual.
- **2. Sparse newer-drug evidence:** some nodes relied on one or a small number of trials.
- **3. Limited long-term safety:** rare harms and outcomes after years of treatment remain uncertain.
- **4. Trial populations differ:** metabolic risk, dose escalation and adherence were not uniform.
- **5. Real-world constraints were outside scope:** cost, supply, access and sustained adherence were not compared.


### Takeaway — Use the ranking as evidence, not a shopping list

The analysis supports a real trend: multi-receptor incretin treatments produced larger average weight-loss estimates than older GLP-1-only drugs. A treatment decision still requires an approved indication, contraindication review, adverse-effect discussion, monitoring, affordability and follow-up with a qualified clinician.


## Primary sources

- [Telegram post 1542](https://t.me/CNSmydream/1542)
- [BMJ Medicine paper](https://doi.org/10.1136/bmjmed-2026-003026)
- [Open full text in PubMed Central](https://pmc.ncbi.nlm.nih.gov/articles/PMC13536023/)
- [PubMed record 42688617](https://pubmed.ncbi.nlm.nih.gov/42688617/)
- [Europe PMC record and full-text index](https://europepmc.org/article/MED/42688617)
- [US FDA label for tirzepatide (Zepbound)](https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22ZEPBOUND%22&limit=1)

---
_Published by iyu (https://iyu.app) — the day's AI news, checked against primary sources and rewritten in plain language. Free to quote with attribution and a link to the canonical URL._