Natural experiments of herpes zoster vaccination: Quantifying the implied vaccine effectiveness against dementia

Recent natural experiments conducted in Wales, Australia, and Canada found a lower risk of all-cause dementia among individuals eligible for live-attenuated herpes zoster (shingles) vaccination than among those who were ineligible, but the extent to which these findings were attributable to vaccination has rarely been quantified. We estimated the reduction in dementia risk among vaccinated individuals implied by the published findings, accounting for incomplete vaccination among eligible individuals and vaccination among ineligible individuals. Combining the reported eligibility effects with differences in vaccine uptake yielded implied reductions in dementia risk among vaccinees of 18% (95% confidence interval -0-35) in Wales, 177% (28-301) in Australia, and 84% (18-219) and 111% (15-303) in the two Canadian analyses. Thus, two estimates exceeded the theoretical maximum effectiveness of 100%, while the third non-Welsh estimate was also implausibly large; even under very high assumed vaccine effectiveness, explaining these findings would require underlying dementia risks among individuals vaccinated because of eligibility to be 1.7-2.3 times higher than in vaccine-ineligible populations, an unlikely scenario. The implied reductions in mild cognitive impairment and dementia-related mortality associated with vaccination were also large, at 41% and 56%, respectively. If vaccination reduced dementia risk by effects of the magnitude implied by these natural experiments, a decline might be expected in population-level dementia incidence following implementation of herpes zoster vaccination programmes; however, no corresponding decline was observed in national data from the UK, Australia, or Canada. In conclusion, the reported natural experiments on herpes zoster vaccination and dementia cannot be considered strong evidence of a clinically meaningful effect; their implied vaccination effects are internally inconsistent and, in some cases, impossible or biologically implausible, highlighting the need for independent replication.