Scientific deep-dive

Do GLP-1 Drugs Reduce Dementia Risk? Three Analyses, Three Answers

In 10 GLP-1 trials, dementia odds were 45% lower, but pooled across all 23 trials reporting dementia the result was not significant. A cohort study found 33% lower risk; a BMJ living review found nothing significant, at low certainty.

By Nora Bissett · Pricing Editor
Editorially reviewed. Not medical advice · How we verify contentLast reviewed
7 min read·3 citations

Three respectable analyses published in 2025 looked at whether these drugs reduce dementia risk, and they do not agree. In a meta-analysis of 26 randomized trials, the 10 that tested GLP-1 drugs showed 45% lower odds of dementia, although the pooled result across the 23 trials that reported dementia was not significant.[1] A cohort study drawn from 396,963 patients with type 2 diabetes found a 33% lower risk.[2] A continuously updated network meta-analysis in the BMJ found no significant effect at all, and graded its own certainty as low.[3] Anyone who has read one headline about this has read a third of the evidence.

Why anyone thinks there might be an effect

Diabetes is an established risk factor for dementia, and the mechanisms proposed to link them (vascular damage, insulin signaling in the brain, chronic inflammation) are all things a glucose-lowering drug might plausibly touch. GLP-1 receptors also exist in the brain, which is why these drugs affect appetite at all, so a direct neurological effect is not far-fetched. The same idea has been raised for epilepsy, where a records study found fewer new cases on GLP-1 drugs, though a pattern inside its numbers points away from neuroprotection.

Plausible mechanism plus observational signal is the combination that has repeatedly failed in dementia research. Hormone replacement therapy, vitamin E and NSAIDs all looked protective in cohort studies and did not survive randomized testing. That history is the reason to read what follows carefully rather than hopefully.

What each analysis found

Three 2025 analyses of GLP-1 drugs and dementia risk.
StudyDesignResult
JAMA Neurology[1]Meta-analysis of 26 randomized trials, 164,531 participantsGLP-1RAs (10 of the 26 trials): OR 0.55 (0.35–0.86)
JAMA Neurology[2]Target trial emulation, 396,963 screened, 33,858 in the GLP-1 comparisonHR 0.67 (0.47–0.96); 2.26 fewer cases per 1,000 person-years
BMJ[3]Living systematic review and network meta-analysisOR 0.92 (0.83–1.02) — not significant, low certainty

The first two are the ones that made the news. The third is the one that keeps being updated, and its authors describe the question as uncertain in as many words.[3]

The disagreement inside the positive results

One detail usually gets lost, and it says more than any single number. The meta-analysis of trials found GLP-1 drugs significantly protective and SGLT2 inhibitors not (OR 1.20, 0.67–2.17), with a formal test confirming the two classes behaved differently.[1] The cohort study found both classes protective, and no difference between them at all (HR 0.97).[2]

If this were a straightforward property of the drug, those two results would agree.

That they do not points at something else. Possibly what is being measured is better diabetes care rather than a specific molecule; possibly dementia diagnoses in health records depend on who is being seen and how often. Both studies are good. The tension between them is a finding in itself.

None of these trials was designed to measure dementia. The randomized trials pooled in the meta-analysis were mostly cardiovascular and kidney outcome trials that happened to record cognitive events; dementia develops over decades and these ran for a few years. A trial that short can rule out a large, fast effect, and any effect on a disease that builds over decades is unlikely to be fast. Where a trial did test thinking directly, the result can be fragile: in a small semaglutide trial in people with HIV, three improved cognitive domains shrank to one after adjustment.

What would settle it

A trial designed for the question, powered for it, and long enough to matter. Those are being run, and they will report years from now. How the dementia signal fits with the death, cancer and muscle data is in our longevity evidence review.

Until then, the BMJ’s verdict stands: uncertain. That is not a satisfying answer, but it is the accurate one, and it is the one to hold against the marketing that has already begun. If someone is selling you a GLP-1 on a dementia claim, they are ahead of three analyses that cannot agree with each other.

For what these drugs demonstrably do across other systems, our beyond-the-scale article sorts the established effects from the associations, and the substance use piece works through a similar pattern of large cohort signals meeting disappointing trials.

Frequently Asked Questions

Nobody knows, and three 2025 analyses disagree. In a meta-analysis of 26 randomized trials, the 10 GLP-1 trials showed 45% lower odds, a cohort drawn from 396,963 patients found 33%, and a continuously updated BMJ network meta-analysis found no significant effect and graded its certainty as low.
They use different designs with different weaknesses, and they even disagree internally: the trial meta-analysis found GLP-1 drugs protective while SGLT2 inhibitors were not, whereas the cohort study found both protective and no difference between them. If this were a simple drug property, those results would align.
No. The pooled randomized trials were mostly cardiovascular and kidney outcome studies that recorded cognitive events along the way, and they ran for a few years. Dementia develops over decades, so these can rule out a large fast effect and little else.
There is no basis for that yet. Plausible mechanism plus observational signal is the combination that failed for hormone replacement therapy, vitamin E and NSAIDs in dementia prevention. Trials designed for the question are running and will report years from now.
None of these analyses found one. The disagreement is about whether there is a benefit, not about whether there is damage.

References

  1. 1.Seminer A, Mulihano A, O'Brien C, et al. Cardioprotective Glucose-Lowering Agents and Dementia Risk: A Systematic Review and Meta-Analysis JAMA Neurology. 2025. PMID: 40193122.
  2. 2.Tang H, Donahoo WT, DeKosky ST, et al. GLP-1RA and SGLT2i Medications for Type 2 Diabetes and Alzheimer Disease and Related Dementias JAMA Neurology. 2025. PMID: 40193118.
  3. 3.Nong K, Jeppesen BT, Shi Q, et al. Medications for adults with type 2 diabetes: a living systematic review and network meta-analysis BMJ. 2025. PMID: 40813122.

GLP-1 Drugs and Hidradenitis Suppurativa: Early Studies, No Trial

Early studies of liraglutide and semaglutide report improvement in HS severity and quality of life. There is no randomized trial, and weight loss by other means has had mixed results in this condition, which complicates the obvious explanation.

5 min read

DSIP: A Sleep Peptide Named in 1977, With No Known Receptor

Delta sleep-inducing peptide was isolated in 1977 and named after a brain-wave effect seen in rabbits. A review of the field states the sleep hypothesis is extremely poorly documented and weak, and that no gene, protein or receptor has been isolated.

4 min read

What We Can and Cannot Verify

Five errors found in our own live pages in one week, all named. Plus the standing limits: 182 unverified LegitScript claims, 324 sellers with no state list, and the things nobody can check from outside.

6 min read

PT-141 (Bremelanotide): What the Evidence Shows

PT-141 is bremelanotide, approved as Vyleesi only for low sexual desire in premenopausal women. Its phase 3 trials found small gains in desire and no rise in satisfying sexual events. The men's program stopped in 2007 over blood pressure.

11 min read

KPV: What the Evidence Shows, and What FDA Found

KPV, a fragment of alpha-MSH, calmed gut inflammation in mouse colitis studies. FDA's 2026 review found no human data. It is not approved or on the 503A bulks list; under an interim HHS policy dated October 2026, HHS does not intend to take enforcement action against topical or oral 503A compounding under strict conditions.

9 min read

Dry Mouth on a GLP-1: Three Published Cases, and What to Tell Your Dentist

The most detailed published evidence that GLP-1 drugs cause dry mouth is three case reports, alongside an adverse-event reporting signal for semaglutide. Nobody has measured how often it happens, but reduced saliva raises decay risk regardless of cause, which is the part worth acting on.

5 min read

Where to get GLP-1 online, safely: sellers our editors have checked

These are telehealth sellers our editors have checked. For each one we hold a price, the form the drug comes in, and the states it reaches.

No insurance needed · vetted by our editors

Some of the links on this page earn us money. If you sign up with a provider after following one, that provider may pay GLP Watchdog a commission. Learn more

8.1

Sesame Care

Oral orforglipron alongside the injectables

6.3

WellMedRx

Comparing compounded and brand GLP-1 plans with a published state list

8.8

altRX

Compounded semaglutide from $89/month at the same price at every dose

About this analysis

GLP Watchdog publishes independently. Each seller in our dataset is checked against that seller's own website, and the dataset is updated continuously. Read our full methodology →