Herpes zoster (shingles) results from reactivation of latent varicella-zoster virus and is targeted by vaccination programmes for older adults. Interest has emerged in whether the recombinant shingles vaccine might also lower risks of cardiovascular outcomes. Prior observational studies comparing vaccinated and unvaccinated people can be biased by the healthy‑vaccinee effect. To reduce such bias, the authors exploited a natural experiment created by the rapid transition in the United States from the live attenuated vaccine to the recombinant vaccine after October 2017.
The study used the step change in vaccine use around October 2017 as a quasi-random exposure. Individuals vaccinated in the six months immediately before the transition (1 April–30 September 2017), when almost all received the live attenuated vaccine, were compared with those vaccinated in the same calendar months one year later (1 April–30 September 2018), when the recombinant vaccine predominated. The identical calendar window a year apart was chosen to remove seasonal confounding in cardiovascular event rates.
The primary analysis matched by propensity score to account for measured covariates and any gradual drift in characteristics of the vaccinated population. After matching, there were 36,460 individuals in the 2018 cohort (93.5% received the recombinant vaccine) and 36,460 in the 2017 cohort (98.6% received the live vaccine). The authors report balance checks and supplementary tables to document covariate distributions and matching performance.
The primary endpoint was a composite of ischemic heart disease, ischemic stroke and heart failure assessed over 7 years after vaccination. The restricted mean time lost (RMTL) ratio was used to compare cumulative burden of the composite endpoint between cohorts, with Kaplan–Meier estimates and piecewise hazard ratios presented for temporal analyses.
Over 7 years, the cohort predominantly receiving the recombinant vaccine had a lower cardiovascular burden than the live vaccine cohort: RMTL ratio = 0.91 (95% CI: 0.88–0.95, P < 0.001), indicating a 9% relative reduction in time lost to the composite outcome and more time lived diagnosis-free. The authors report an E value of 1.42, indicating the magnitude of unmeasured confounding that would be required to fully explain the observed association after matching.
The association held in both sexes and did not show evidence of moderation by sex (permutation test P = 0.53).
Analyses of individual endpoints showed significant reductions in burden for ischemic heart disease (RMTL ratio = 0.90, 95% CI: 0.87–0.94) and heart failure (RMTL ratio = 0.88, 95% CI: 0.83–0.93). Ischemic stroke showed a significant reduction in males (RMTL ratio = 0.88, 95% CI: 0.78–0.98). Atrial fibrillation incidence was also lower (RMTL = 0.93, 95% CI: 0.88–0.98).
No association was observed for myocarditis, peripheral arterial disease, hemorrhagic stroke or transient ischemic attack. The association for ST-elevation myocardial infarction (STEMI) had a similar point estimate to ischemic heart disease and heart failure but did not reach statistical significance.
The protective association against ischemic heart disease and heart failure was concentrated in the first half of follow-up. Kaplan–Meier differences and piecewise hazard ratios indicate hazard ratios below 1 in the first ~3.5 years, with attenuation and non-significant hazard ratios in the latter half of the 7-year follow-up, leading to plateauing of cumulative-incidence differences over time.
Multiple sensitivity and secondary analyses supported the robustness of the findings. Results were consistent when accounting for death as a competing risk and there was no association with non-cardiovascular death. Findings persisted when restricting follow-up entirely to the pre-COVID-19 period, when follow-up time was matched between cohorts, and when different time windows were used to ascertain covariates. In the year before vaccination, the two cohorts did not differ in preventive healthcare use or in prior cardiovascular risk. Preventive-health indicators during follow-up were also similar between cohorts.
As negative-control checks, those who received the tetanus, diphtheria and pertussis (TDaP) vaccine in April–September 2018 had a very similar risk of recorded cardiovascular events to those who received TDaP in the same months in 2017, arguing against secular diagnostic trends as an explanation. The recombinant vaccine group, as expected, had lower shingles incidence than the live vaccine group and similar risk for a negative-control outcome.
Using a natural-experiment design and propensity-score matching, the investigators observed that receipt of the recombinant shingles vaccine was associated with a modest but statistically significant reduction in the 7-year cardiovascular burden among adults aged ≥60, driven by reductions in ischemic heart disease and heart failure and with some evidence for lower ischemic stroke risk in males and lower atrial fibrillation incidence.
The association attenuated over time and was strongest in early follow-up. The authors emphasise that while the design mitigates common biases in vaccine observational studies, these are still observational findings. They conclude that the results justify randomized clinical trials and mechanistic studies to investigate whether shingles vaccines have true cardioprotective effects and to understand potential biological mechanisms.