A vaccine switch created a natural experiment on shingles shots and heart disease

Studies claiming that the shingles vaccine protects against something other than shingles share a structural weakness: they compare people who got vaccinated with people who did not, and those two groups differ in every way that predicts health. A paper published in Nature Medicine on August 26 tries to design that problem away [s1].

Its device is the rapid transition in the United States from the live attenuated shingles vaccine to the recombinant one. Adults vaccinated immediately before the switch and adults vaccinated immediately after are, in principle, the same kind of person — people who show up for a shingles vaccine — separated only by which product was in the fridge that month [s1].

What was compared

The authors compared incidences of a composite cardiovascular endpoint — ischaemic heart disease, heart failure or ischaemic stroke — among adults aged 60 years and over vaccinated immediately before versus immediately after the transition [s1].

The effect measure is a restricted mean time lost ratio, which compares the time each group spent having experienced the outcome over the follow-up window rather than a simple hazard. It is a burden measure, and the paper's language reflects that: a decrease in burden, not a decrease in risk of ever having an event.

The results

The recombinant vaccine was associated with a 9% decrease in cardiovascular burden over seven years (RMTL ratio 0.91, 95% CI 0.88–0.95) [s1].

Broken out, the association was significant for ischaemic heart disease — a 10% decrease in burden (RMTL ratio 0.90, 95% CI 0.87–0.94) — and for heart failure, a 12% decrease (RMTL ratio 0.88, 95% CI 0.83–0.93), in both sexes [s1]. For ischaemic stroke it was significant in males only, at a 12% decrease (RMTL ratio 0.88, 95% CI 0.78–0.98) [s1]. Atrial fibrillation showed a 7% decrease in burden (RMTL 0.93, 95% CI 0.88–0.98), while other cardiac, peripheral and cerebrovascular outcomes did not [s1].

The authors note the association attenuated over time [s1] — the gap between the groups was largest early and narrowed across the seven years.

What the design does and does not buy

A natural experiment of this kind removes the crudest form of confounding: healthy-user bias between vaccinated and unvaccinated people. It does not remove everything. People vaccinated in different calendar periods encountered different background medicine, and the two products differ in dose schedule and reactogenicity in ways that could influence who completes a course.

Nor does the composite endpoint escape the usual objection to composites. Ischaemic heart disease, heart failure and stroke are grouped because they are all cardiovascular, and the individual components moved by different amounts.

The authors' own conclusion is appropriately narrow: these results justify clinical trials and mechanistic studies to investigate potential cardioprotective effects of shingles vaccines [s1]. That is a call for evidence, not a claim to have produced it.

The pattern this belongs to

This is the second organ system for which the recombinant shingles vaccine has been credited with a benefit beyond zoster, and the dementia literature is a useful calibration.

A systematic review and meta-analysis published on August 10 pooled 14 reports representing 13 studies of herpes zoster vaccination and dementia [s2]. In its primary analysis of five independent data sources, vaccination was associated with lower dementia risk — a ratio estimate of 0.72 (95% CI 0.65–0.81) — but with I-squared of 97.1%, meaning the studies disagreed with one another almost completely [s2]. Two regression-discontinuity studies reported absolute reductions in dementia diagnosis of 1.3 and 1.8 percentage points tied to vaccination eligibility [s2].

Notably, two direct comparisons favoured the recombinant vaccine over the live attenuated one (RMTL ratio 0.83, 95% CI 0.80–0.87; risk ratio 0.82, 95% CI 0.69–0.98) — the same contrast, and one of the same statistical measures, as the cardiovascular paper [s2].

The review's own verdict is the one to hold onto: substantial heterogeneity and residual confounding preclude causal interpretation, and further prospective or randomised evidence is needed [s2]. Associations that were generally attenuated when an active comparator was used [s2] are a warning that some of the effect in the wider literature is comparison-group artefact.

What to watch

Whether any of the proposed trials are actually run. A randomised comparison powered for cardiovascular endpoints in a vaccinated population would be large and long, and there is no regulatory requirement forcing one. Until then this remains an association from a well-designed observational study, which is a real thing and not the same thing as a cardioprotective effect.

This article is informational and is not medical advice. Vaccination decisions belong with a clinician who knows the individual case.

Sources

• Recombinant shingles vaccination and the risk of cardiovascular events — Nature Medicine, 2026-08-26
• Association Between Herpes Zoster Vaccination and Dementia Risk: A Systematic Review and Meta-Analysis — Vaccines, 2026-08-10