The shingles vaccine and dementia risk: the evidence base as of August 2026

People who receive the shingles vaccine keep turning up with a lower risk of being diagnosed with dementia later. This started as one striking result — a 2025 Nature study exploiting a quirk in how Wales rolled out its vaccination programme — and has since been echoed by a near-identical natural experiment in Australia, a US Medicare cohort of roughly 1.5 million people, and a June 2026 study of more than half a million older adults leaving skilled-nursing care. The estimates cluster somewhere between 20% and 33% lower risk. That is a real and unusually consistent signal. It is also, still, an association: not one of these studies is a randomized controlled trial, and the biggest of them cannot fully rule out the possibility that healthier people are simply more likely to get vaccinated. The first large randomized trial began recruiting in Denmark in April 2026 and will not report until around 2029. Until then, the honest framing is that this is the most promising unproven lead in dementia prevention — and the honest next step is to ask your doctor whether the shingles vaccine is right for you.

Does the shingles vaccine lower dementia risk?

Possibly, and the case has strengthened considerably since 2025. Natural experiments in Wales and Australia, plus very large cohort studies in the US, all point the same way: vaccination is associated with roughly 20–33% fewer dementia diagnoses over follow-up periods of four to nine years. None of it proves cause and effect. The vaccine's established job is preventing shingles.

What changed since this article was first published

When we first wrote this piece, the story rested mostly on one paper. It no longer does. Here is what the field looks like in August 2026, roughly in order of how much weight each piece of evidence deserves.

1. The natural experiments — the strongest evidence, and why

Most "linked to lower risk" headlines share a fatal weakness: the people who did the thing were different to begin with. People who get vaccinated tend to be more health-conscious, see doctors more often, and be better off — all of which independently affect dementia risk. Researchers call this healthy-vaccinee bias, and it is the single biggest reason to distrust vaccine-and-dementia findings.

The 2025 study led by Markus Eyting, Pascal Geldsetzer and colleagues, published in Nature on 2 April 2025, sidesteps that problem in an elegant way. When Wales launched its shingles vaccination programme on 1 September 2013, eligibility was set by a sharp cutoff: adults born on or after 2 September 1933 could get the vaccine; those born even a day earlier could not. People born a week either side of that line are, on average, essentially identical — same health-consciousness, same access to care, same everything. The only meaningful difference is which side of an administrative line their birthday fell on.

Analysing 282,541 older adults who were free of dementia when the programme began, the team found that being eligible for the vaccine was associated with a 3.5 percentage point absolute reduction in new dementia diagnoses over seven years (95% CI 0.6–7.1, p = 0.019) — a 20% relative reduction. The effect was noticeably larger in women than in men.

Three weeks later, the same group published a near-replication in JAMA. When Australia launched its own programme on 1 November 2016, it drew an equally sharp line: adults aged 70–79 on that date got the free live vaccine; those who had already turned 80 did not. Across 101,219 patients in Australian primary care records, Pomirchy and colleagues found vaccine eligibility was associated with a 1.8 percentage point reduction in new dementia diagnoses over 7.4 years (95% CI 0.4–3.3, p = 0.01).

Two different countries, two different health systems, two independent accidental experiments, same direction. That convergence is the most persuasive single fact in this literature.

2. The Wales follow-up: does it affect people who already have dementia?

In December 2025, the Stanford group extended the Welsh natural experiment across the whole arc of the disease, publishing in Cell (Xie, Eyting, Bommer, Ahmed and Geldsetzer). Two findings stand out. Among 282,557 people without cognitive impairment at baseline, vaccine eligibility was associated with a 1.5 percentage point reduction in new mild-cognitive-impairment diagnoses over nine years (95% CI 0.5–2.9, p = 0.006). And among 14,350 people who already had a dementia diagnosis when the programme began — of whom 7,049 died of dementia during follow-up — eligibility was associated with an 8.5 percentage point reduction in dementia deaths (95% CI 0.6–18.5, p = 0.036).

That second result is the one that generated headlines about the vaccine "slowing" dementia. Read the confidence interval before you get excited: the lower bound is 0.6 percentage points, which is close to nothing. The finding is compatible with a large effect and also with a barely detectable one. It is a genuinely interesting hypothesis-generating result, not a demonstration that the vaccine changes the course of anyone's disease.

3. The very large US cohorts — bigger numbers, weaker design

Two enormous American studies arrived in the last year. Both are worth knowing about, and both are more vulnerable to confounding than the natural experiments.

• The Medicare cohort (about 1.5 million people). Published in Alzheimer's & Dementia in December 2025 by dosReis and colleagues, this compared 502,845 Medicare beneficiaries aged 65+ who received both doses of the recombinant vaccine (Shingrix) with 1,005,690 matched unvaccinated comparators. Vaccination was associated with a 33% lower rate of dementia diagnoses (HR 0.67, 95% CI 0.66–0.68), a 27% lower rate for Alzheimer's disease specifically (HR 0.73), and 33% lower for vascular dementia (HR 0.67). The authors tried to blunt healthy-user bias by drawing comparators from people who had attended preventive-care visits. Several co-authors are employees of the vaccine's manufacturer.
• The skilled-nursing cohort (June 2026). Published in Annals of Internal Medicine by Kaleen Hayes and colleagues at Brown University and the University of Delaware, this followed 509,926 adults aged 66 and over admitted to more than 5,500 US skilled-nursing facilities between 2017 and 2022, using a target-trial-emulation design. Over four years, 18.8% of those who received at least one dose of the recombinant vaccine were diagnosed with dementia, versus 24.6% of the unvaccinated — about 24% lower risk. Only 8,843 of the half-million participants were actually vaccinated, which is a very small exposed group relative to the cohort. The authors were unusually candid: vaccinated residents were younger and healthier than those who were not, and the researchers state they cannot be certain the vaccine was the reason for the difference. The study was funded by the vaccine's manufacturer.

Pool it all together and you get roughly the same picture. A systematic review and meta-analysis in Age and Ageing (Maggi and colleagues, November 2025) combined eight cohorts covering 979,768 vaccinated and 9,455,017 unvaccinated people and reported a pooled 24% lower risk of any dementia (RR 0.76, 95% CI 0.69–0.83). The authors flagged high heterogeneity between studies and explicitly named healthy-vaccinee bias as an unresolved limitation.

4. Why might a shingles vaccine affect the brain?

This is where honesty matters most: nobody is certain yet. There are two leading explanations, they are not mutually exclusive, and researchers are now openly arguing about which one is right.

• Calming the virus and the inflammation it triggers. Shingles is caused by the varicella-zoster virus — the same one behind chickenpox — which stays dormant in the body for life and can reactivate later. A large Nature Medicine analysis published in October 2025 examined health records from more than 100 million people in the US and reported that shingles reactivation was consistently associated with higher subsequent dementia risk after adjusting for close to 400 measured characteristics, with recurrent episodes carrying more risk than a single one. If reactivation drives risk, a vaccine that suppresses reactivation is an obvious candidate for lowering it.
• The immune response itself, not the virus. In June 2025, Taquet, Todd and Harrison published a study in npj Vaccines covering 436,788 people, finding that both the shingles vaccine and the RSV vaccine were associated with more dementia-free time than the flu vaccine — 18% and 29% more respectively over 18 months. Since the two share the AS01 adjuvant, the authors proposed that the adjuvant itself might be doing the work. That claim did not go unchallenged: a rebuttal in the same journal in December 2025 (Williams and colleagues, all affiliated with Pfizer, which makes a non-adjuvanted RSV vaccine) argued that around a quarter of the supposed AS01 group had probably received a non-adjuvanted RSV vaccine, that there was no dose-response relationship despite the two vaccines containing different AS01 doses, and that many unrelated vaccine types show similar dementia associations — which would point to something much more general than AS01.

Both mechanisms are hypotheses. That the specialists are publicly disagreeing about which is right should tell you how unsettled this is.

5. The randomized trial has finally started

Every study above shares the same ceiling: none of them randomly assigned anyone to anything. That gap matters, and it isn't hypothetical — GLP-1 drugs looked dramatically protective in health records and then showed nothing in the EVOKE Alzheimer's trials. That is changing here too. DAN-ZOSTER, registered in March 2026, is a nationwide pragmatic randomized trial in Denmark that plans to assign roughly 162,000 adults aged 65 and over in a 1:1 ratio to either two doses of the recombinant shingles vaccine or no vaccine. Its two primary outcomes are major adverse cardiovascular events and incident dementia. Recruitment began on 28 April 2026; primary completion is estimated for April 2029.

Notably, that trial tests the recombinant vaccine — while most of the strongest evidence, the Wales and Australia natural experiments, concerns the older live-attenuated vaccine (Zostavax). Geldsetzer argued in Nature Medicine in 2026 that a large randomized trial of the live-attenuated vaccine is urgently needed too, and pointed at an awkward structural problem: that vaccine is now off-patent, so no company has a commercial reason to fund the study.

The honest caveats — read these before you draw conclusions

The volume of evidence has grown a lot. The strength of the inference has grown much less. Keep these front of mind:

• Still no completed randomized trial. A natural experiment is powerful but it is not the gold standard, and everything else here is observational. A definitive cause-and-effect link is not established, and will not be until roughly 2029 at the earliest.
• Healthy-vaccinee bias is only partly handled. The Wales and Australia designs largely neutralise it. The million-person US cohorts do not — they can only adjust for what they happened to measure. When a design that controls confounding well (Wales, ~20%) and designs that control it poorly (US cohorts, ~24–33%) disagree in magnitude, the more conservative number is usually the one to trust.
• Under-diagnosis distorts everything. Both natural experiments flagged substantial under-recording of dementia in routine records. In the Australian data, only about 1.4% of the sample had a recorded diagnosis against an estimated true prevalence above 8% in that age group.
• The estimates apply to a narrow age band. The regression-discontinuity results are driven overwhelmingly by people around 79–80 years old at the cutoff. They do not tell you what happens if you are vaccinated at 65.
• Confidence intervals are wide. The headline numbers are point estimates. The Welsh absolute effect ranges from 0.6 to 7.1 percentage points; the dementia-mortality result runs from 0.6 to 18.5. The true effect may be much smaller than the headline.
• Manufacturer involvement is common. The Medicare cohort, the 2026 skilled-nursing study and the varicella-zoster reactivation analysis in Nature Medicine were all funded by or co-authored with GSK, which makes Shingrix — and the rebuttal to the AS01 hypothesis was written by authors affiliated with Pfizer, which makes a competing RSV vaccine. This is normal in vaccine research and does not make any of the results wrong. It is context you deserve, and it cuts both ways.
• "Associated with lower risk" is not "prevents." The vaccine is not proven to prevent dementia, and it is not a treatment for anyone who already has it.
• The vaccine's actual job is preventing shingles. Shingles can be painful and, in older adults, seriously debilitating. That — not brain health — is why it is recommended. Any dementia signal is best thought of as a potential bonus.
• Solenna is not recommending any vaccine. We don't give medical advice, and we're not telling anyone to get — or skip — a vaccine. Whether the shingles vaccine is appropriate for you depends on your age, health and history, and that's a conversation for your doctor.

So what should you actually do with this?

If you're an older adult, the practical takeaway hasn't changed with all this new evidence: talk to your doctor about whether the shingles vaccine is right for you. For many eligible people it is already recommended to prevent shingles itself, and this research adds an interesting extra reason to have the conversation. Your clinician can weigh your specific situation, including which vaccine and what timing make sense.

It's also worth keeping the finding in proportion. Even at its most optimistic, a vaccine is one factor among many. The 2024 Lancet Commission estimated that around 45% of dementia cases worldwide are associated with 14 modifiable risk factors — things like blood pressure, hearing, physical activity, blood sugar, sleep, smoking, mood and social connection. Those levers are available to you today, they compound over time, and consistency is what makes them count. The shingles vaccine is a promising piece of an emerging puzzle; the everyday factors are the well-established foundation. The smartest move is to pay attention to both — ask your doctor about the vaccine, and keep working the levers you can control.

How we research and review our content →

This article is educational and is not medical advice, diagnosis, or treatment. Studies cited — including the 2025 Nature and JAMA natural experiments, the US Medicare and skilled-nursing cohorts, and related analyses — describe associations and risk factors in general populations, not Solenna specifically, and Solenna does not recommend for or against any vaccine or medication. Solenna does not diagnose, prevent, treat, or cure Alzheimer's disease or any form of dementia; individual results vary and no outcome is guaranteed. Do not start, stop, or change any vaccine or medication without speaking to a qualified healthcare professional about your own situation.