Neurology
Causal evidence that herpes zoster vaccination prevents or delays a proportion of dementia cases: A natural experiment in Wales Felix Michalik* Pascal Geldsetzer Markus Eyting Min Xie Simon Heß
Aims: To determine the causal effect of herpes zoster (HZ) vaccination on new diagnoses of dementia.
Methods: In the UK, eligibility for the HZ vaccine (Zostavax) was determined based on an individual’s exact date of birth. Those born before September-2-1933 were ineligible (and remained ineligible for life), while those born on or after September-2-1933 were eligible. This natural experiment likely provides causal, as opposed to correlational, evidence because there is no plausible reason why adults born just one week apart should differ systematically from each other. We used a regression discontinuity design in Welsh country-wide data on primary and secondary care encounters, and death certificates. The outcome was a new dementia diagnosis as recorded in any of these data sources.
Result: We included 282,541 individuals born between September-1-1925 and August-31-1941. The percentage who received the HZ vaccine increased from 0.01% among those one week too old to be eligible, to 47.2% among those one week younger. Receiving the vaccine reduced the probability of a new dementia diagnosis over seven years by 3.5 percentage points (95% CI: 0.6–7.1, p=0.019), corresponding to a 22.4% relative decrease. To support causality, we show that i) there were no differences in pre-existing conditions or uptake of other preventive interventions across the date-of-birth eligibility cutoff; ii) the HZ vaccine had no effects on any other common causes of morbidity and mortality; and iii) no other interventions used the same date-of-birth eligibility cutoff.
Conclusions: Our findings suggest that HZ vaccination slows or prevents the natural history of dementia. Unlike existing studies in this field, this analysis provides causal evidence because individuals who differ in age by just one week are likely exchangeable with each other on both observed and unobserved characteristics, except for a large difference in the probability of receiving the HZ vaccine.
