Older adults who get the shingles vaccine are diagnosed with dementia less often. That finding has been replicated in three separate countries, using a method sturdier than ordinary observational studies. Even so, it does not count as proof that the vaccine prevents dementia: no randomized controlled trial has tested the question, and an independent recalculation published in 2026 showed that the reported effect comes out impossibly large in some countries.
Shingles appears when the virus behind childhood chickenpox (varicella zoster) reawakens in the nerves years later. The vaccine was developed to prevent that reawakening. A link to dementia was not something researchers went looking for; the data raised the question on their own.
Comparing vaccinated and unvaccinated people directly is misleading. People who get vaccinated tend to see their doctor more often and take better care of their health; if they develop dementia less often, the vaccine could be the reason, but so could their lifestyle. Researchers call this "healthy vaccinee bias."
Wales offered a way to untangle this knot. In the vaccination program that began on September 1, 2013, eligibility was set by date of birth: people born before September 2, 1933 never became eligible for the vaccine at any point in their lives, while those born on or after that date could get vaccinated for at least a year. Among people born one week before the cutoff, the vaccination rate was 0.01%; among those born one week after, it was 47.2%. There is no reason for these two groups to differ systematically in education, diet or health habits. The only difference between them is the right to the vaccine that the calendar handed them.
The team of Markus Eyting and Pascal Geldsetzer at Stanford compared the health records of 282,541 people on either side of this cutoff and published the results in Nature in 2025. Over seven years of follow-up, 35,307 people received a new dementia diagnosis. Actually receiving the vaccine lowered the probability of a new dementia diagnosis by 3.5 percentage points, which corresponds to a 20% relative reduction (95% confidence interval 6.5–33.4%). The team found the same pattern in death certificates from England and Wales, for deaths due to dementia.
A result from a single country could be a fluke. So the same team looked at other countries where similar date-of-birth rules applied.
In Australia, free vaccination began on November 1, 2016 for people aged 70–79; those born just after the cutoff were 16.4 percentage points more likely to be vaccinated. The two groups were balanced in their past illnesses and in their use of other preventive health services, and eligibility for the vaccine did not affect diagnoses of common chronic conditions other than dementia. The Canadian study went a step further: after the program began, dementia diagnoses in the relevant birth years in Ontario were less frequent than in the same age groups in other provinces that had no shingles vaccination program.
A follow-up analysis published in Cell in December 2025 applied the Welsh data to both ends of the disease. Among 282,557 people with no record of cognitive impairment at baseline, vaccination lowered diagnoses of mild cognitive impairment by 3.1 percentage points over nine years. And among the 14,350 people who already had a dementia diagnosis when the program began, deaths due to dementia were 29.5 percentage points lower in those who were vaccinated.
The second figure is striking, but its confidence interval runs from 0.6 to 62.9. In other words, the true effect could be very close to zero, or very large. The authors' reading that "the vaccine may be slowing the course of the disease" should be treated as a hypothesis for now.
All of the natural experiments described so far were carried out with the older, live attenuated vaccine (Zostavax). Many countries have dropped that vaccine and switched to the recombinant vaccine (Shingrix). Maxime Taquet's team at Oxford took advantage of that rapid switch in the United States after October 2017: most people vaccinated just before the switch had received the live vaccine, and most of those vaccinated just after had received the recombinant one.
In the analysis, based on the records of more than 200,000 people and published in Nature Medicine in 2024, the recombinant vaccine was associated with 17% more time lived without a diagnosis over six years. For those who later went on to be diagnosed with dementia, that means 164 additional diagnosis-free days. Dementia risk was also lower in people who received the recombinant vaccine than in those who received the flu vaccine or the tetanus-diphtheria-pertussis vaccine. The authors state plainly that the study is observational and cannot show causation.
Another analysis, published in Nature Medicine in October 2025, screened the records of more than 100 million people in the United States. Recurrent shingles was associated with a higher dementia risk than a single episode, while shingles vaccines were associated with a lower risk than the pneumococcal vaccine chosen as the comparison. Most of this study's authors are employees of GSK, the maker of Shingrix; that is worth keeping in mind when reading the results.
There is no definitive answer. A review published in Clinical Infectious Diseases in October 2026 highlights three pathways that are not mutually exclusive:
For Shingrix there is a fourth candidate as well: the immune stimulant (adjuvant) in the vaccine, called AS01. In 2025, Taquet and colleagues reported that in a matched cohort of 436,788 people, both the AS01-containing shingles vaccine and the RSV vaccine carrying the same adjuvant were associated with a reduction in dementia risk over 18 months. No difference emerged between the two vaccines. This suggests that part of the effect may come from the adjuvant and not from the virus; however, 18 months is a very short follow-up period for dementia.
In the Welsh study, the protective effect was markedly larger in women than in men. Yet the vaccine's power to prevent shingles and the jump in the vaccination rate were similar in both sexes. In the Cell follow-up analysis, too, the reductions in mild cognitive impairment and in deaths due to dementia were larger in women. In the Shingrix comparison, women gained 22% more diagnosis-free time and men 13%.
Researchers are weighing two possibilities. The off-target immune effects of vaccines have often been found to be stronger in women in other studies as well. On the other hand, because dementia is less common in men at advanced ages, the confidence interval around the estimates remains wide; a meaningful benefit in men cannot be ruled out either.
In September 2026, the team of Mika Kivimäki and Gill Livingston at University College London recalculated the published figures from the natural experiments. The question they asked was this: given that only some of those who became eligible actually got vaccinated, how large a risk reduction among the vaccinated does the reported difference imply? For Wales the answer was 18%. For Australia it was 177%, and for the two analyses in Canada, 84% and 111%.
It is not possible to reduce a risk by more than 100%. The authors also note that national data from the United Kingdom, Australia and Canada show no comparable drop in how common dementia is after the vaccination programs. Their conclusion is that these natural experiments cannot be considered strong evidence of a clinically meaningful effect and need to be replicated by independent teams. The debate continued through an exchange of letters in the August 2026 issue of Lancet Neurology.
The two systematic reviews from the same year are cautious as well. A review by the Norwegian Institute of Public Health covering 18 primary studies confirmed, with high certainty of evidence, that the recombinant vaccine is about 95% effective against shingles in randomized trials; for the reduction in dementia risk, however, it graded the certainty of evidence as "low" and pointed out that similar associations are seen with other vaccines too. Another meta-analysis, which screened 13 studies, found a 28% lower dementia risk among the vaccinated in its main analysis pooling five independent data sources (ratio 0.72; 95% confidence interval 0.65–0.81), but wrote that no causal interpretation is possible because of the large inconsistency between studies and residual confounding.
The proven benefit of the shingles vaccine is preventing shingles and the nerve pain that can follow it. It has no approved use for preventing dementia. Whether the vaccine is right for you depends on your age, your immune status and the medications you take; make that decision in conversation with your doctor. Treating the dementia findings as the sole reason to get vaccinated does not match today's evidence; seeing them as a possible added benefit is reasonable.
While the vaccine debate continues, you can also look at other topics that matter for the aging brain and body: our articles on sleep and longevity and strength training weigh the evidence on those topics by the same yardstick.
That has not been proven. In natural experiments in Wales, Australia and Canada, dementia diagnoses were less common in the age groups that became eligible for the shingles vaccine; the Welsh study calculated a 20% relative reduction over 7 years among those vaccinated. But there is no randomized controlled trial on the question, and an independent assessment published in 2026 showed that some of the reported effects are too large to be plausible.
An ordinary observational study compares people who were vaccinated with people who were not; because the vaccinated tend to be people who take better care of their health, the result can be misleading. In a natural experiment, a rule the person did not choose, such as date of birth, determines who becomes eligible for the vaccine. People born a week before and a week after the cutoff are very similar; the only systematic difference between them is access to the vaccine.
In a 2024 study based on US health records, people who received the recombinant vaccine (Shingrix) lived 17% more time without a dementia diagnosis over 6 years than those who received the live vaccine; among those who were later diagnosed, that corresponds to 164 additional days. Because the study is observational, it does not show causation. All of the natural experiments, meanwhile, were carried out with the older live vaccine.
It is not known. In the Welsh study the protective effect was markedly larger in women than in men; in the Shingrix comparison, too, diagnosis-free time was 22% longer in women and 13% longer in men. Researchers point out that the off-target immune effects of vaccines are often stronger in women, but add that because dementia is less common in men, the estimates remain more uncertain.
The shingles vaccine was developed to prevent shingles and the nerve pain that follows it; it has no approved use for preventing dementia. Whether the vaccine is right for you depends on your age, your immune status and the medications you take. Make this decision together with your doctor; viewing the dementia findings as a possible added benefit is the approach that best fits the current level of evidence.
"Natural experiments based on date of birth have delivered the strongest data in this field. But the fact that the same data imply an effect above 100% in some countries shows the story is not over yet."
— Vivelong Longevity Science Team