In 2023, a Stanford researcher named Pascal Geldsetzer was examining the way Wales had rolled out the shingles vaccine a decade earlier — and noticed something peculiar.
To ration the limited supply, Welsh health authorities had drawn a line in the sand: if you were 79 on September 1, 2013, you qualified. If you'd already turned 80, you didn't [1].
Two people, born a week apart. Same town. Same GP. Same risk factors. One could get the vaccine. The other never could. And because Wales has detailed health records on essentially everyone, researchers could follow them for the next seven years and compare what happened.
That's about as close to a randomised trial as you ever get without actually running one. What Geldsetzer found in those records was striking enough that it ended up published in Nature, then replicated in Australia, then Canada. Earlier this year, it landed him on the TIME 100 list of the most influential people in health and medicine [2].
Table of Contents
- The New Cardiovascular Evidence
- The Healthy-Vaccinee Problem
- Why Wales Is Different
- The Bridge to the Vaccine You'll Actually Get
- What This Means for You
- Closing Thoughts
- References
This article examines something newer, and in some ways more unexpected. Because if Geldsetzer was right that the shingles vaccine is doing something helpful to the brain, the obvious next question is: Is it helping anywhere else? In the past year, three separate studies — including one tracking more than a million people for up to eight years — have come back with the same answer.
The shingles vaccine is associated with a substantially lower risk of heart attacks, strokes, and heart failure.
The dementia side of Geldsetzer's work was covered in a previous article — but the cardiovascular finding is new, it's bigger, and the question worth examining here is whether it's actually real.
Because that headline sounds too good. Free heart protection from a shot you're already being told to get? That should make every well-trained sceptic uneasy. This article walks through three questions: How good is the evidence, really? What's the most plausible biological explanation? And — most importantly — what should adults aged 50 or older actually do with this information?
The New Cardiovascular Evidence
If shingles itself can damage your blood vessels — and it can — then a vaccine that stops shingles from reactivating should plausibly do something to vascular outcomes. That's not a stretch.
The shingles virus is a herpesvirus. After a chickenpox infection in childhood, it lies dormant inside the nervous system for the rest of a person's life. When it reactivates — which is what causes shingles — it doesn't just damage the nerve it travels down. It can also infect the walls of blood vessels, triggering inflammation, vascular damage, and in some cases, stroke or heart attack downstream [3].
So the question stops being "why would a shingles vaccine touch the heart?" and becomes "how big is the effect, and can the studies that try to measure it actually be trusted?"
Three observational studies have come out in quick succession [4][5][6]. The strongest was led by a team in South Korea. Published in the European Heart Journal, the study followed more than 1.27 million adults aged 50 and older for a median of six years. Researchers used a statistical technique called propensity-score overlap weighting to make the vaccinated and unvaccinated groups look as similar as possible at baseline [4].
The results: those who received the vaccination had a 23% lower risk of cardiovascular events overall. Heart failure, down 26%. Major adverse events — heart attack, stroke, or death from heart disease — also down 26%. And the protective signal lasted up to eight years after vaccination [4].
That's the cleanest piece of evidence available. A million people. Eight years of follow-up. Peer-reviewed. And — important caveat — it was the older live-attenuated version of the vaccine, not the one most adults in 2026 will actually be offered. More on that below.
The second study was a systematic review, pooling nine studies. The numbers were more conservative — a 16% to 18% reduction in cardiovascular events — but the results pointed in the same direction [5].
But there's a problem with all the data reviewed so far. The third study illustrates it clearly. It was a retrospective look at almost a quarter of a million US adults with established cardiovascular disease. The vaccinated group had a 46% lower risk of major adverse cardiac events in the first year post-vaccination [6].
That sounds enormous — and it should raise eyebrows. The same study reported a 66% reduction in all-cause deaths in the first year after a single vaccination. These numbers are far larger than what's typically seen even with established cardiovascular therapies [6].
That 66% mortality figure isn't just impressive — it's implausibly large. The numbers are far larger than what's typically seen even with established cardiovascular therapies. And that discrepancy points toward a well-known methodological issue in the underlying data.
The Healthy-Vaccinee Problem
People who choose to get a non-mandatory vaccine are not a random slice of the population. They're more likely to see their GP regularly. They're more likely to take their medications. They tend to have higher socio-economic status, lower smoking rates, and better preventive-care engagement across the board.
In epidemiology, this is called healthy user bias. And it's the standard alternative explanation for any observational study showing a vaccine protects you against something the vaccine wasn't designed for.
The three studies covered above — the Korean one, the ESC meta-analysis, the ACC abstract — all share the same fundamental weakness.
But here's where it gets interesting. If healthy-user bias were the whole story, you'd expect the benefit to show up most clearly in the healthiest people — the ones whose baseline behaviours are already driving the effect. In the Korean study, the opposite happened.
The people who benefited most from vaccination were the ones with unhealthy baseline behaviours [4].
That's hard to square with a pure selection effect. If vaccination were just a proxy for being healthy, the unhealthy subgroup is exactly where you'd expect the signal to wash out. Instead, it's where the signal was strongest.
That doesn't settle the question. Selection effects can hide inside subgroups — the smoker who gets vaccinated may still be more engaged with healthcare than the smoker who doesn't. But it does introduce a note of caution about the "this is all just the healthy user bias" explanation.
To establish whether there's a true causal relationship here — rather than a statistical artefact of who chooses to get vaccinated — a randomized controlled trial would be needed. And that's where the natural-experiment design that put Geldsetzer on the TIME 100 becomes directly relevant.
Why Wales Is Different
Remember the line in the sand with the vaccine rollout in Wales at the start of this article?
Born before the 2nd of September 1933 — never eligible. Born on or after — eligible.
That cutoff didn't care about health behaviours. It didn't care whether someone saw their GP. It didn't care whether they smoked or exercised. It cared about a single date on a birth certificate. And in the dementia analysis Geldsetzer's team published in Nature last year, eligibility for the vaccine produced a substantial protective effect on dementia diagnoses — about a 20% relative reduction [7].
That dementia finding — roughly a 20% relative reduction in diagnoses, with a clean causal design — was covered in detail in a previous article.
The point that matters here is structural, not clinical. If the vaccine produces a real biological effect on a major outcome in a study where healthy user bias has been engineered out by design, that lends weight to the cardiovascular signal showing up in the messier observational studies. Different organ. Different research teams. Different countries. Different datasets. Same direction. That consistency, across independent designs, is worth taking seriously — even in the absence of a purpose-built RCT.
The Bridge to the Vaccine You'll Actually Get
There's a crucial distinction to flag here. Almost everything described above — the Korean study, the Welsh natural experiment, most of the dementia work — used the older shingles vaccine, called Zostavax. A live, weakened virus.
That vaccine has been retired in most of the world. CDC guidelines in the U.S., for instance, specify the recombinant vaccine, Shingrix. The older Zostavax was phased out in 2020 [8].
Shingrix is a fundamentally different product. It has a different antigen — just a piece of the virus, not the whole thing — and a powerful adjuvant called AS01 designed to provoke a stronger immune response. A fair question is: do the off-target benefits seen with Zostavax carry over to the vaccine most adults will be offered today?
The honest answer is: probably yes, but the evidence here is one notch weaker.
An Oxford team led by Maxime Taquet used the moment in 2017 when the United States switched from Zostavax to Shingrix as a quasi-natural experiment. They compared roughly 100,000 Shingrix recipients to roughly 100,000 Zostavax recipients, matched on baseline characteristics, and followed them for six years. Shingrix was associated with about 17% more dementia-free time than Zostavax — meaning, in this dataset, Shingrix looked at least as protective, and possibly more so [9].
Two caveats on that. First, this is observational again. Second, one of the senior authors has a consulting relationship with GSK, the manufacturer of Shingrix. The research team states that GSK had no role in this specific study — but it's the kind of disclosure that should accompany any reading of the findings [9].
What This Means for You
For adults aged 50 or older who have not been vaccinated against shingles, the case for getting Shingrix has gotten materially stronger. The original reason — preventing a painful, sometimes disabling rash and the lingering nerve pain (post-herpetic neuralgia) that can follow it for months — was already a reasonable one. The dementia signal added weight. The cardiovascular signal adds more. None of these is yet proven by a purpose-built RCT, but the convergence of signals across independent datasets, countries, and research teams is increasingly hard to dismiss.
In the United States, the CDC recommends two doses of Shingrix, two to six months apart, for all adults aged 50 and up with normal immune function [8].
One thing GPs may not always emphasise enough: Shingrix is, to put it politely, a vaccine you'll feel. Injection-site soreness, fatigue, headache, and sometimes fever are common after both doses. Roughly one in six recipients had reactions severe enough to interfere with daily activities [10].
That's not an emergency — it's the immune system doing what the adjuvant is designed to make it do — but it's worth knowing in advance. Practical tip: scheduling the second dose for a Friday afternoon means the worst of the reaction lands on a Saturday rather than in the middle of a workweek.
For those who already received the older Zostavax vaccine years ago, the CDC recommends completing the two-dose Shingrix course anyway [8].
Shingrix is more effective at preventing shingles itself, and the off-target evidence for the recombinant vaccine looks at least as strong [11].
What the current evidence does not support is seeking out the vaccine outside standard eligibility. The evidence base is strong enough to be another reason not to skip Shingrix when the turn comes. It is not yet strong enough to recommend off-label vaccination of 40-year-olds. There's a real difference, and the distinction is worth preserving.
Closing Thoughts
There's one detail about Geldsetzer's work worth noting as a closing thought. When he first noticed the Welsh birthdate cutoff in those records, he wasn't looking for a vaccine that lowered dementia. He was looking at how policy decisions get made when supplies are limited. The dementia finding came as a surprise to him. The cardiovascular finding came as a surprise to a different team in Korea, looking at a completely different population. Sometimes the most useful evidence in medicine doesn't come from the trial that was designed to find it.
It comes from a date on a birth certificate. From a propensity-weighted spreadsheet of a million people in a country whose government tracks vaccinations against insurance claims. From an accident.
References
1. https://med.stanford.edu/news/all-news/2025/03/shingles-vaccination-dementia.html
2. https://time.com/collections/time100-health-2026/7362629/pascal-geldsetzer/
3. https://pmc.ncbi.nlm.nih.gov/articles/PMC5298244/
4. https://academic.oup.com/eurheartj/article/46/30/2991/8124786
5. https://academic.oup.com/eurheartj/article/46/Supplement_1/ehaf784.3633/8312335
7. https://www.nature.com/articles/s41586-025-08800-x
8. https://www.cdc.gov/shingles/hcp/vaccine-considerations/index.html
9. https://www.nature.com/articles/s41591-024-03201-5
10. https://www.cdc.gov/vaccine-safety/vaccines/shingles-herpes.html
