The shingles vaccine keeps getting linked to lower dementia risk, but linked is not the same as causes
A large 2026 study of more than half a million older adults in US nursing homes found that people who got the Shingrix shingles vaccine were diagnosed with dementia less often than those who did not, 18.8 percent versus 24.6 percent over four years. The finding has circulated widely again in late August, and it is genuinely interesting. It is also observational, which means it can show a strong association but cannot by itself prove the vaccine is the cause. Only a small fraction of the group was vaccinated, vaccinated people were slightly healthier to begin with, and the researchers themselves say clinical trials are needed before anyone claims the shot prevents dementia. This is a report on what the study found and what it cannot yet show, not medical advice.
A study published this year in Annals of Internal Medicine, led by researchers at Brown University, analysed more than 509,000 older adults in skilled nursing facilities and found that those who received the Shingrix shingles vaccine developed dementia at a lower rate than those who did not: an estimated 18.8 percent versus 24.6 percent over a four-year follow-up, a 24 percent relative reduction. The researchers describe it as roughly one in 17 dementia cases potentially avoided, if the effect turns out to be causal. The result drew renewed attention in late August 2026. The crucial caveat is built into the study's design: it is observational, so it can establish a strong association but not prove the vaccine causes the lower risk. Only 8,843 of the participants were vaccinated, vaccinated people tended to be younger and healthier, and adjusting for that did not fully remove the gap. The authors say clinical trials are needed. This is a science report, not medical advice; decisions about vaccination belong with your doctor.
Every so often a striking health finding goes around again, and the shingles vaccine's apparent link to lower dementia risk is one of the stickiest. A large study out of Brown University, published in 2026, put a hard number on it, and that number, a roughly one-quarter reduction in dementia diagnoses among vaccinated older adults, is easy to repeat and easy to overstate. The finding is real and worth taking seriously. The reason to slow down is not that it is weak, but that the type of study it is can only take the claim so far.
What did the study find?
According to Brown University and the report published in June 2026 in Annals of Internal Medicine, researchers examined more than 509,000 adults aged 66 and over who were admitted to over 5,500 skilled nursing facilities across the United States between 2017 and 2022, using a combination of Medicare claims and electronic health records. To be included, people had to have no prior dementia diagnosis and to be eligible for the shingles vaccine.
Over a four-year follow-up, as ScienceDaily summarised, an estimated 18.8 percent of vaccinated participants went on to be diagnosed with dementia, compared with 24.6 percent of those who were not vaccinated. Those percentages are model-derived estimates from the statistical method the study used, rather than raw head counts, but every summary reports them the same way. That is a 24 percent relative reduction and an absolute difference of about 5.8 percentage points, which the researchers translate into roughly one in 17 dementia cases potentially being prevented, if the effect is real and causal. The vaccine in question is the current recombinant shingles vaccine, Shingrix (RZV), given as one or two doses.
Why "observational" changes how you should read it
This is the heart of the matter. The study is observational: the researchers did not randomly assign people to get the vaccine or not, they looked back at what happened to people who, in real life, did or did not get it. That design is powerful for spotting associations across huge populations, and target trial emulation, the method used here, is a careful way to squeeze causal-style comparisons out of such data. But it cannot fully rule out the possibility that something other than the vaccine explains the gap.
The most important confounder is who gets vaccinated. As the researchers acknowledge, vaccinated people "tended to be slightly younger and healthier" than those who were not. People who keep up with preventive care like vaccines also tend to exercise more, attend more medical appointments and manage other conditions better, all of which independently affect dementia risk. This is the "healthy-vaccinee" effect, and it can create exactly this kind of association without the vaccine doing the work. The team adjusted their analysis for measurable differences, and importantly they report that doing so "did not fully explain the association," which means simple, measured gaps in age and health do not obviously account for the difference. But statistical adjustment can only correct for things that were measured, so unmeasured factors, the exercise, appointment-keeping and better management of other conditions that tend to travel with getting vaccinated, could still be doing some of the work. It is not the same as the clean comparison a randomised trial provides.
One more disclosure belongs on the table. The study was funded by GlaxoSmithKline, the company that makes Shingrix. The authors state that the funder had no role in the study's design, analysis or the decision to publish, and industry funding does not by itself make a result wrong. But when a finding could help sell a product, who paid for the work is exactly the kind of context a careful reader should be given, so it is worth saying plainly.
How big is the vaccinated group, really?
Small, relative to the headline. Of the more than 509,000 people in the study, only 8,843 were vaccinated, under two percent. A large total sample sounds reassuring, but the comparison that actually drives the result rests on that much smaller vaccinated subgroup, drawn from a specific population: older adults in skilled nursing facilities. That is a frailer, more medically monitored group than the general older population, so the findings do not automatically transfer to a healthy 66-year-old living independently. None of this makes the result wrong. It just means the number describes a particular comparison in a particular setting, not a universal law.
Is this the first time we have seen this link?
No, and that context matters. Studies have suggested a connection between shingles vaccination and lower dementia risk before, but as study author Kaley Hayes put it, "a lot of previous studies with similar results focused on an older vaccine." The best-known prior evidence came from the previous-generation live shingles vaccine, and some of it used natural-experiment designs based on eligibility cut-off dates, which are harder to confound. This study's contribution is to find a similar signal for the current Shingrix vaccine in a large US population. So the honest framing is not "a new study proves the vaccine stops dementia," but "the association keeps showing up, now including the vaccine people actually receive today, across different methods and populations." A repeated association across studies is more persuasive than any single one, and it is still not proof of cause.
What would it take to actually prove it?
A randomised controlled trial, which is precisely what the authors call for. In that design, people would be assigned to receive the vaccine or not without regard to how healthy they are, removing the healthy-vaccinee problem, and the dementia rates would be compared directly. Until something like that is done, the responsible statement is the one the researchers themselves make: they cannot say for certain that the vaccine was the reason vaccinated adults developed dementia less often. The mechanism people speculate about, that suppressing viral reactivation might reduce inflammation or vascular damage in the brain, is plausible and interesting, but a plausible mechanism plus a strong association is a hypothesis worth testing, not a settled conclusion.
The study at a glance
| Published | 2026, Annals of Internal Medicine (Brown University and collaborators) |
| Design | Observational, target trial emulation on Medicare claims plus health records |
| Population | 509,926 adults aged 66+ in 5,500+ skilled nursing facilities, 2017 to 2022 |
| Vaccinated | 8,843 (about 1.7 percent), with the current Shingrix (RZV) vaccine |
| Main result | Dementia in 18.8% of vaccinated vs 24.6% of unvaccinated over four years |
| Effect size | 24% relative reduction; about one in 17 cases potentially prevented, if the link is causal |
| Key limitation | Cannot prove causation; vaccinated people were slightly younger and healthier |
| Authors' caveat | Adjustment did not fully explain the gap, but clinical trials are needed |
The right response to a finding like this is neither to dismiss it nor to treat it as proven. A consistent, sizeable association across large populations is a strong reason to run the trials that could settle the question, and it is a genuinely encouraging signal for an outcome, dementia, where prevention options are scarce. It is not, yet, a reason to say the shingles vaccine prevents dementia, because the studies that could show that have not been done. Holding both of those thoughts at once is the whole skill, the same care we bring to early results in Alzheimer's blood testing and other fast-moving corners of medicine, where the gap between promising and proven is exactly where the science happens.
The FDA just approved the first mRNA flu shot, a year to the day after Washington cut off federal mRNA vaccine funding
On 5 August 2026 the FDA approved mFlusiva, Moderna's mRNA influenza vaccine and the first of its kind for seasonal flu. It arrived exactly one year after the government cancelled roughly $500 million in mRNA vaccine work, and only after the agency's own vaccine chief moved to block it, then reversed. Here is what the shot does, and the fight behind it.
Webb's spectra point to a hidden crowd of small stars in the early universe, and it makes the first big galaxies harder to explain
In a study published on 18 August in Nature Astronomy, a Leiden University team combined deep James Webb Space Telescope spectra with earlier ground-based data for nine massive galaxies that had already stopped forming stars in the young universe. The spectra imply these galaxies hold a far larger share of faint, low-mass stars than a Milky-Way-like assumption predicts. That hidden mass makes the galaxies heavier than earlier estimates, by amounts that vary across the sample, with one standout perhaps up to four times heavier. The authors argue this deepens rather than solves the puzzle of why the early universe already held such massive galaxies.
An mRNA cancer vaccine reported its first positive Phase 3 result. Here is what that does and does not mean
On 19 August, Merck and Moderna said their personalized mRNA vaccine intismeran autogene, given with Keytruda, met its main goals in a 1,137-patient Phase 3 melanoma trial: it lengthened the time patients stayed cancer-free after surgery and the time before the disease spread. It is the first positive Phase 3 for any individualized-neoantigen or mRNA cancer therapy. But the companies released only top-line results, no efficacy numbers yet, it is not approved, and it treats a cancer already removed by surgery rather than curing advanced disease.