In a powerful natural experiment using Australian health data, researchers found that being eligible for the shingles vaccine may reduce dementia diagnoses, strengthening the case for preventive strategies in brain health.
Image Credit: BlurryMe / Shutterstock
In a recent study published in the JAMA (The Journal of the American Medical Association), an international team of researchers determined whether eligibility for herpes zoster (HZ) (virus that causes shingles) vaccination based on date of birth influenced the likelihood of receiving a new dementia (memory loss and thinking problems) diagnosis.
Background
Dementia affects over 55 million people globally, posing a growing public health crisis. While age remains the strongest risk factor, infections may also play a role. One underexplored link is between HZ and dementia. HZ results from reactivation of the varicella zoster virus, a neurotropic virus that can affect the central nervous system. Vaccination against HZ might not only prevent shingles but also lower dementia risk, possibly through immune modulation. A previous quasi-experiment in Wales found this association, but replication is essential across diverse populations and health systems. Further research is needed to validate these findings globally.
About the study
The present study employed a quasi-experimental design using primary care data from 65 general practices across Australia, facilitated by the health informatics platform PenCS. The analysis leveraged a natural eligibility cutoff created by the National Immunisation Programme, which began offering the live attenuated HZ vaccine (Zostavax) free of charge on November 1, 2016, to individuals aged 70 to 79 years. Eligibility was determined by birthdate: individuals born on or after November 2, 1936, were eligible, while those born before were not. This setup allowed for a comparison between groups that were nearly identical in age and baseline health, differing primarily in vaccine access.
Patient records included diagnosis histories, immunizations, prescriptions, and demographic details. Dates of birth were coded by week, and all diagnoses, including dementia, were identified using open-ended text fields provided by general practitioners. Patients aged 50 or older as of November 1, 2016, and with at least one clinical visit between 1993 and 2024 were included.
The primary outcome was the first recorded diagnosis of dementia during a 7.4-year follow-up period. The main exposure was eligibility for HZ vaccination based on birthdate. Statistical analysis centered on regression discontinuity (RD), comparing individuals born just before and after the eligibility threshold. This method controls for both observed and unobserved variables, assuming no abrupt changes other than vaccination status. Secondary analyses used time-to-event models, including accelerated failure time and Kaplan-Meier survival analyses, along with robustness checks across multiple bandwidths and modeling strategies. All analyses were conducted using R statistical software.
It is important to note that the effect measured in this study is for eligibility for HZ vaccination, not for confirmed receipt of the vaccine, because vaccination status is likely underreported in the primary care data used. Due to this underreporting, the study authors did not attempt to estimate the effect of actually receiving the vaccine, as this could overstate the results.
Additionally, the study population was drawn from practices that agreed to participate and use the PenCS platform, so the data are not fully representative of all Australian primary care patients. The effect estimate is also "local," applying most clearly to individuals who were around 79 to 80 years old at the time the HZ vaccination program began.
The protective effect observed in this study specifically pertains to the live attenuated HZ vaccine (Zostavax), as the newer recombinant vaccine (Shingrix) was not widely used in Australia during the study period.
Study results
Data from 101,219 patients were analyzed, focusing on 18,402 patients born within 482 weeks of the November 2, 1936, eligibility threshold. The mean age in this subset was 77 years, with 54.3% of the participants being women. The probability of receiving the HZ vaccine jumped from 6.5% among ineligible individuals to 30.2% among eligible individuals, confirming that the date-of-birth rule effectively differentiated vaccine exposure.
Importantly, no differences were observed in prior health conditions, uptake of other preventive services, or dementia risk factors across the eligibility threshold, supporting the validity of the natural experiment. Regression discontinuity analysis showed that eligibility for HZ vaccination led to a statistically significant 1.8 percentage point reduction in the probability of receiving a new dementia diagnosis over 7.4 years (95% confidence interval: 0.4 to 3.3; P = .01). The protective effect was consistent across alternative follow-up durations, grace periods, and model specifications.
Additional checks, including those limited to frequent primary care users and time-to-event models, supported the primary findings. No effects were observed on other common diagnoses or preventive health behaviors, indicating the result was specific to dementia. Comparative RD using an additional ineligible cohort yielded a similar effect size of 1.5 percentage points. Kaplan-Meier plots and cumulative incidence curves further showed the delayed onset of dementia in vaccine-eligible individuals.
The study ruled out confounding by confirming that no other interventions used the same date-of-birth eligibility rule and demonstrating that the effect was unique to the 1936 birth threshold. Analyses shifting the threshold to nearby years showed no similar effect, further validating the causal interpretation.
It is also important to note that dementia diagnoses are substantially underdetected in the primary care data analyzed. For example, only about 1.4% of patients older than 65 in the PenCS dataset had a dementia diagnosis, compared to an estimated 8.4% prevalence in the general Australian population. This underdiagnosis means that the absolute effect size observed may not fully reflect the impact of HZ vaccination on dementia risk in the broader population.
These findings, when combined with similar research from Wales, provide consistent and compelling evidence that HZ vaccination may help prevent or delay the onset of dementia.
Conclusions
To summarize, this study demonstrated that individuals eligible for free HZ vaccination due to their date of birth had a significantly lower likelihood of being diagnosed with dementia over a 7.4-year follow-up period. The use of a quasi-experimental design allowed for a comparison between nearly identical groups, minimizing confounding and providing stronger causal inference than traditional observational studies. These findings highlight the potential for HZ vaccination to serve as a low-cost intervention for dementia prevention. However, further studies in additional populations, as well as mechanistic and clinical research, are needed to explore the biological pathways, generalizability, and policy implications of these promising results.