Does the shingles vaccine lower dementia risk?
Claim attributed to Academic researchers reporting natural-experiment studies (Eyting et al. 2025 Nature; Taquet et al. 2024 Nature Medicine), amplified by longevity media. , The strongest finding comes from a Stanford/Welsh-data team funded by the US NIH with no competing interests, not from a vaccine seller. Corroborating Oxford cohort work is NIHR-funded but has a GSK-consultant co-author.
A near-randomized natural experiment found shingles vaccination lowered new dementia diagnoses by roughly 20% relative, the strongest evidence yet for this kind of question. It is supported but not yet proven prevention: no trial has been run with a dementia endpoint.
A near-randomized birth-date cutoff says vaccination lowers dementia diagnoses by about a fifth, but no trial has tested it directly and the effect may not even be specific to the shingles virus.
What it’s supposed to target
- Varicella-zoster virus (VZV) reactivation
- Neuroinflammation
- Cerebrovascular damage (VZV vasculopathy)
- Trained / non-specific immunity
Two mechanisms are proposed. The specific one: the vaccine suppresses reactivation of varicella-zoster virus (VZV), the chickenpox virus that lurks in nerves for life. Reactivation can trigger neuroinflammation and VZV vasculopathy (inflamed, damaged brain blood vessels), both plausibly feeding dementia, so blocking reactivation may protect the brain. The non-specific one: vaccines, especially adjuvanted ones, can produce trained immunity, broad immune changes that might lower dementia risk independent of VZV. The chain: vaccination → less viral reactivation and a recalibrated immune system → less of the inflammation and vascular injury linked to cognitive decline.
What makes this stand out is not the mechanism (still uncertain) but the design behind the evidence: a natural experiment exploiting a vaccine-eligibility birthdate cutoff approximates a randomized trial, far stronger than the usual observational fare, and it sidesteps the bias that healthier people get vaccinated. Even so, the pathway is not pinned down, the effect may differ by sex and by vaccine type (live Zostavax versus recombinant Shingrix), and a dedicated randomized trial with a dementia endpoint is still pending. A strong quasi-experimental signal with a plausible mechanism, not a closed case.
Mechanism is theory, not proof. A plausible pathway explains why something might work, not whether it does. The verdict rests on the evidence below, not the elegance of the theory.
What would have to be true
Vaccination must reduce dementia diagnoses (holds: ~20% relative, ~3.5-point absolute reduction in the Welsh experiment).
The association must be causal, not confounding (largely holds: the birth-date cutoff mimics randomization and avoids healthy-vaccinee bias).
The mechanism must be understood (open: VZV-suppression versus a non-specific AS01 adjuvant effect is not yet disentangled).
What the evidence actually shows
A birth-date cutoff that mimics a trial
Wales offered the live Zostavax only to people born on or after 2 September 1933, so vaccine receipt jumped from near zero to about 47% across a single week of birth dates. Eyting et al. (2025, Nature) exploited this as a regression-discontinuity instrument and found eligibility cut new dementia diagnoses by 3.5 percentage points (95% CI 0.6 to 7.1) over seven years, a roughly 20% relative reduction, stronger in women. Because the cutoff is essentially arbitrary, the design approximates a randomized trial and avoids the healthy-vaccinee bias that weakened earlier observational work.
Corroboration, and a hint the virus may not be the point
An Oxford natural experiment (Taquet et al. 2024, Nature Medicine, ~208,000 people) found the recombinant Shingrix was linked to lower dementia risk than live Zostavax (about 17% more diagnosis-free time), again larger in women. But a 2025 NPJ Vaccines cohort found an unrelated AS01-adjuvanted RSV vaccine carried a similar signal, suggesting part of the benefit may be a non-specific adjuvant effect rather than shingles-specific. Crucially, no randomized trial has yet tested a dementia endpoint.
Studies, graded, and who paid
A regression-discontinuity design that approximates a trial found a 3.5-point absolute, about 20% relative, reduction over 7 years.
The date-of-birth cutoff sidesteps healthy-vaccinee bias, but no randomized trial has confirmed it.
An AS01-adjuvanted RSV vaccine showed a similar signal, hinting at a non-specific adjuvant effect.
| # | Study | Type | Size | Funding / COI | Key limitations |
|---|---|---|---|---|---|
| 1 | Eyting et al., Nature 2025 (Welsh natural experiment) | Quasi-experimental regression-discontinuity on EHR data | ~56,098 adults | Independent US NIH grants DP2 AI171011 and R01 AG084535; no competing interests. | Studies the now-discontinued live Zostavax; modest absolute effect; 7-year follow-up; sex difference unexplained. |
| 2 | Taquet et al., Nature Medicine 2024 (live-to-recombinant switch) | Propensity-matched natural-experiment cohort | ~207,674 (two matched cohorts) | Mixed UK NIHR funded; co-author J.A. Todd is a GSK consultant; paper states GSK had no involvement. GSK makes Shingrix. | Observational comparison, not randomized; residual confounding possible; GSK-linked author. |
| 3 | Taquet et al., NPJ Vaccines 2025 (AS01 shingles vs RSV) | Propensity-matched cohort (US EHR) | ~436,788 | Mixed UK NIHR funded; Todd is a GSK consultant and Oxford-GSK co-director; GSK no involvement. AS01 is GSK's adjuvant. | Only 18-month follow-up; suggests effect may be non-specific; confounding possible. |
| 4 | Nature Medicine 2026 commentary (call for RCTs) | Expert commentary | N/A | Funding unknown Full text paywalled; existence, title and thrust confirmed via DOI and search. | Cited only for the RCT-pending framing, not as primary evidence. |
| 5 | University of Oxford press release 2024 | Institutional secondary summary | ~200,000 (describing the primary study) | , University communications office describing the NIHR-funded Taquet study. | Secondary summary; adds no independent evidence. |
Multiple independent teams using different natural experiments converge on the same direction, which is unusually consistent for an observational question.
Unproven ≠ disproven
The benefit was discovered serendipitously, so it rests on quasi-experiments rather than a pre-registered trial; durability beyond 6 to 7 years is untested.
Where claim and evidence diverge
The strongest causal evidence is for the live Zostavax, now largely replaced by Shingrix, whose dementia benefit rests on weaker observational comparisons.
The money trail
The pivotal Welsh result is NIH-funded and independent of vaccine makers. The Oxford cohort work is NIHR-funded but has a GSK-consultant co-author; GSK makes Shingrix and the AS01 adjuvant, though it had no role in the studies.
The honest read
Take the shingles vaccine for shingles; a meaningful drop in dementia risk is a plausible and well-supported bonus, but it is not yet proven and the absolute benefit is modest.
What would change this verdict
A randomized trial with a dementia endpoint showing no benefit would downgrade this.
Evidence that the birth-date cutoff was confounded by some other 1933-era exposure would weaken the causal claim.
Sources
- Eyting M, Xie M, Michalik F, et al. A natural experiment on the effect of herpes zoster vaccination on dementia. Nature. 2025;641(8062):438-446. PMID 40175543.
- Taquet M, Dercon Q, Todd JA, Harrison PJ. The recombinant shingles vaccine is associated with lower risk of dementia. Nature Medicine. 2024;30(10):2777-2781. PMID 39053634.
- Taquet M, Todd JA, Harrison PJ. Lower risk of dementia with AS01-adjuvanted vaccination against shingles and RSV infections. NPJ Vaccines. 2025;10:130. PMID 40562756.
- Why large-scale randomized trials of live-attenuated shingles vaccination for dementia prevention are urgently needed. Nature Medicine. 2026.
- University of Oxford. New shingles vaccine could reduce risk of dementia. Oxford News, 25 Jul 2024.
- Pomirchy M, et al. Herpes zoster vaccination and incident dementia in Canada: an analysis of natural experiments. Lancet Neurol. 2026. PMID 41579903.
- Xie M, et al. The effect of shingles vaccination at different stages of the dementia disease course. Cell. 2025. PMID 41338191.
People also ask
- Does the shingles vaccine reduce the risk of dementia?
- The evidence supports it but it is not yet proven. A near-randomized Welsh natural experiment (Eyting et al., 2025) found vaccine eligibility cut new dementia diagnoses by about 3.5 percentage points over seven years, a roughly 20% relative reduction, stronger in women.
- How did researchers study the shingles vaccine and dementia without a trial?
- Wales offered the live Zostavax only to people born on or after 2 September 1933, so vaccine receipt jumped sharply across a single week of birth dates. Researchers used this arbitrary cutoff as a regression-discontinuity design that approximates a randomized trial and avoids healthy-vaccinee bias.
- Is Shingrix or Zostavax better for lowering dementia risk?
- The strongest causal evidence is for the live Zostavax, now largely replaced by Shingrix. An Oxford natural experiment (Taquet et al., 2024) linked recombinant Shingrix to about 17% more dementia-free time than Zostavax, but that rests on weaker observational comparisons.
- Is the dementia benefit specific to the shingles virus?
- Not clearly. A 2025 cohort found an unrelated AS01-adjuvanted RSV vaccine carried a similar signal, suggesting part of the benefit may be a non-specific adjuvant effect rather than shingles-specific. No randomized trial has yet tested a dementia endpoint.
Caveat is journalism, not medical advice. We check public claims against published evidence; we don’t diagnose, treat, or tell you what to take.