The new study sounds like the beginning of a much larger story. Researchers at Oxford, analyzing the electronic health records of 72,000 vaccinated Americans followed for seven years, found that people who received the newer shingles vaccine, Shingrix, had 9 percent less cardiovascular disease than those who received the older one, Zostavax: 10 percent less coronary heart disease, 12 percent less heart failure, and 12 percent fewer strokes in men. The results were published in Nature Medicine, and they arrive on top of earlier work by the same team suggesting the same vaccine is associated with less dementia. A vaccine that protects the heart and the brain would be one of the biggest findings in preventive medicine in decades.
The reason to take the study seriously is also the reason to keep the celebration on hold, and it is the same thing: the design. The researchers found a clever way to imitate a randomized trial, and the imitation is both the study's strength and its ceiling. Everything that makes the result believable is borrowed from a policy decision made in 2017, and everything that keeps it from being proof comes from the fact that the imitation is still not the real thing.
The design was borrowed from a regulator's calendar
The American shingles vaccine market changed overnight in October 2017, when regulators approved Shingrix and it quickly displaced Zostavax. Suddenly, people who went to get vaccinated received one vaccine, and people who had gone a few months earlier received the other. The Oxford team built its study on that boundary. It took people over sixty who were vaccinated within six months on either side of the approval, matched them by age, sex, and health history, and compared the two groups across seven years of medical records from 60 health systems.
That is why the design is better than an ordinary observational study. The classic criticism of vaccine studies is the healthy vaccinee bias: people who get vaccinated tend to be healthier, wealthier, and more engaged with medical care, so they have better outcomes for reasons that have nothing to do with the vaccine. A comparison of vaccinated against unvaccinated people is a comparison of two different populations wearing the same label. In this study, everyone was vaccinated. Everyone walked into a clinic and asked for a shingles shot. The only difference is which shot the calendar handed them, a difference no patient controlled and no doctor chose for medical reasons. That single fact eliminates the most obvious confounder, which is why the study deserves more weight than the usual association.
The design also explains why this study arrived now. The boundary needs time to produce outcomes. A vaccine that prevents heart disease works over years, not months, so the comparison only becomes informative once both cohorts have accumulated enough follow-up for events to accrue. Seven years after October 2017 is roughly the earliest moment a study like this could ask its question with statistical power, which is why the same team's dementia papers, and now the cardiovascular one, have landed in sequence rather than at once.
The numbers are real and modest
The findings are consistent across the cardiovascular categories, with the composite burden of heart attacks, heart failure, and stroke 9 percent lower in the Shingrix group over seven years, and atrial fibrillation 7 percent lower. The stroke benefit showed up clearly in men and not in women, which the authors cannot yet explain. The absolute difference is the number that travels least well: about 1 percent fewer cardiovascular events over seven years. As Mark Russell, a King's College London researcher commenting for the Science Media Centre, put it, the benefit is "relatively small on an individual basis," though it could add up to a substantial number of prevented events across the tens of millions of people who receive the vaccine.
The restraint of the researchers is notable, and it is part of the evidence. Maxime Taquet, the Oxford psychiatrist who leads the work, said plainly that the study is "observational, even though it's a natural experiment," and that the right response for now is to get the vaccine for its actual purpose, shingles prevention, and treat any heart or brain benefit as an unproven bonus. The mechanism is unknown. One hypothesis is that the chickenpox virus, which stays dormant in the body and reawakens as shingles, quietly raises cardiovascular risk, so a stronger vaccine means less risk. The other is that Shingrix's distinctive ingredient, an immune-boosting adjuvant called AS01, retrains the immune system itself. The fact that Zostavax, which also prevents shingles, shows no cardiovascular benefit in the data points toward the adjuvant, but both stories are speculation at this point.
The ceiling is built into the design
The same cleverness that removed the healthy vaccinee problem leaves other problems standing. The people on the two sides of the October 2017 boundary chose to be vaccinated at different times, and timing correlates with everything from flu season to insurance coverage to how a person reacts to news about a new vaccine. The records come from health systems that serve privately insured patients, leaving out the large population on public coverage, a hole the authors acknowledge. Matching on recorded health history cannot equalize everything that is not recorded. And seven years of records cannot distinguish a vaccine effect from the thousand small differences in behavior and circumstance that separate people who got one shot from people who got the other.
None of this is a criticism the authors would dispute. It is the price of the design. A natural experiment is an imitation, and the imitation can only do what a calendar does. A real randomized trial assigns the vaccine by coin flip, and the world's largest such trial is now running in Denmark for exactly this question.
The unexplained subgroup findings point the same way. The stroke benefit appeared in men and not in women, the atrial fibrillation effect is smaller than the others, and no one can say yet whether those patterns are biology or noise. A clean causal story would not leave such loose ends; a real effect can, and the honest reading is that the loose ends are exactly what a randomized trial is for. Independent experts who reviewed the study, including Brown University's Kaleen Hayes, called the cardiovascular question the obvious next one after the dementia findings and noted that the prior observational evidence on the heart had been conflicting, which is another way of saying the field needed a well-designed observational study precisely because the randomized answer did not exist yet.
Denmark will settle what Oxford can only suggest
The Danish trial, called DAN-ZOSTER, is the answer to the ceiling. It is randomizing roughly 162,000 people aged sixty-five and older to receive Shingrix or no vaccine, then reading their outcomes from Denmark's national health registries. More than a million Danes were invited by mail to join. The primary results for cardiovascular events are expected next year, 2027, with dementia outcomes following later, and the vaccine's maker, GSK, is co-sponsoring with the Danish research institutions. It is open-label, and it cannot be otherwise, since a placebo shot would be impractical at this scale, but the coin flip is real. The trial's design is pragmatic in the best sense: the intervention is the ordinary two-dose schedule, the outcome data already exist in registries, and the question is the same one the Oxford team posed, stripped of the imitation. If the 9 percent effect is real, the trial is large enough to find it. If it is not, the trial is large enough to bury it.
That trial will resolve the ambiguity in a way the Oxford team has been careful to flag all along. The observational trail behind Shingrix is unusually deep: the same group's 2024 study of veterans found 17 percent longer time free of dementia among Shingrix recipients, a larger 2025 analysis reported a 27 percent lower dementia risk with two doses, and a natural experiment in Wales found the older vaccine reduced dementia diagnoses by 20 percent. When independent designs keep pointing the same direction, the signal gets harder to dismiss, but the entire trail is still built from observation, and observational signals in vaccine science have a long record of shrinking or vanishing when tested randomly.
The heart study, then, is best read as a beautifully constructed question rather than an answer. It raised the stakes of the Danish trial from interesting to urgent, because a randomized confirmation would mean that a vaccine already recommended to most older adults is quietly preventing heart attacks and strokes too, at a scale few medicines achieve. Until 2027, the correct posture is the one the researchers themselves maintain: get the shot for the shingles protection it was approved for, and let Denmark decide whether the rest of the benefits are real. The good news in the meantime is that nobody has to choose. The vaccine is already worth taking on its own terms, and if the calendar in Oxford turns out to have been right about the heart and the brain as well, it will have been the rare medical discovery that arrived as a bonus to people who were protected all along.
Primary sources
- The STAT report by Elizabeth Cooney for the new Nature Medicine study's design and results, the quotes from Maxime Taquet, Mark Russell, and Kaleen Hayes, and the mechanism hypotheses.
- The DAN-ZOSTER registration on ClinicalTrials.gov for the Danish trial's design, size, endpoints, and timeline.
- The CDC's shingles vaccination guidance for the vaccine's approved purpose and recommendation.