More Evidence That The Shingles Vaccine Reduces Dementia Risk

New data from Australia show a benefit in both women AND men

It’s only been a few weeks since a fascinating study showed that getting the shingles vaccine reduced the risk of dementia in older adults in Wales.

Hot on its heels, a new study published in JAMA used a similar natural experiment in Australia and reassuringly found very similar results. (I should note that the papers have the same corresponding author, , Dr. Pascal Geldsetzer, but the rest of the authorship team is different).

To re-cap the findings of the first paper published in Nature, people born just before Sept 2nd, 1933 in Wales never received a shingles vaccine, compared to a sizeable portion of people born just after the eligibility cut-off. Being on the lucky side of the birthday cut-off and getting a shingles vaccine led to a 20% lower relative risk of a dementia diagnosis over the next 3 years.

The authors worked painstakingly to rule out other explanations for their results, which I found quite convincing. The only head-scratcher was that the effect was seen only in women, with the effect for men estimated at close to zero. While there may be sex differences in innate immune responses linked with dementia that could explain this result, we’d obviously prefer if such an easy and effective intervention to prevent dementia worked for the whole population (we want to save the brains of men too!). Women have an increased risk of developing dementia over their lifetime compared to men, but it’s not completely clear whether this is a real elevated risk or a result of more women living to the oldest ages when dementia is most common. But if women were at higher risk biologically, it could make sense that they might benefit more from an intervention that impacts one of those mechanisms.

Another “Natural Experiment”

The new study used a very similar design, taking advantage of the roll-out of the shingles vaccine in Australia as a natural experiment. In this case, the eligibility birthday cut-off was November 2, 1936, but the set-up was very similar. As a reminder, this type of “regression discontinuity design”, or RDD, mimics an experiment by assuming that people born right before or right after this cut-off are very similar (on average), except for their access to this vaccine. So, the “treatment” of getting the shingles vaccine could almost be thought of as determined by a coin flip, by whether one’s mother went into labor one day earlier or later. If this “quasi-random” assignment is true, an RDD design is quite compelling for causal inference.

Source: Pomirchy M, Bommer C, Pradella F, Michalik F, Peters R, Geldsetzer P. Herpes Zoster Vaccination and Dementia Occurrence. JAMA. Efigure 1, Supplementary Material

Above you can see that the chances of getting the shingles (herpes zoster or HZ) vaccine jump a lot (are “discontinuous”) around the birthday cut-off, as we would hope for the experimental design. But the chances of getting the flu vaccine do not look different around the threshold (panel B), which is reassuring. If lots of other things changed around this birthday threshold, we’d be suspicious that there was some broader change in health services or other policies going on, and less confident that we could attribute any changes in dementia to the shingles vaccine alone. But if the shingles vaccine is the only thing that changes, we’re more confident in our causal conclusions.

Results: The Shingles Vaccine Prevents Dementia Again

Over 7.4 years of follow-up, the study found that people eligible for the shingles vaccine had a significantly lower incidence of dementia; 3.7% of the eligible group were diagnosed over follow-up compared to 5.5% of those not eligible for the vaccine.

How does the size of this effect compare to the Wales study? Unfortunately, it’s a bit tricky to make a direct comparison. While the study from Wales covered most of the population thanks to linked records with the National Health Service, the Australian data came from a private company offering primary care data from only 65 general practitioner offices around Australia. This data may not be as representative of the whole population and GP offices don’t capture all health events as comprehensively as the NHS. The authors suggest that both vaccination and dementia were likely underestimated since national estimates of levels from other sources are higher. This shouldn’t bias the identification of a causal effect of the vaccine on dementia from the quasi-experimental design, but it does make it harder to interpret the size of the effect. Being eligible for the vaccine reduced the relative risk of dementia by 33% in this sample (1.8%/5.5%). Since not everyone who was eligible got the vaccine, this should mean that the effect of actually getting the vaccine is even stronger. But since both vaccination and dementia were likely underreported, the authors weren’t comfortable trying to scale up the estimate for actual vaccine take-up as they did in the Wales paper where the records were better. Still, for me these effect sizes bolster the takeaway from the previous paper that this effect is large enough to be substantively important.

As you can see below, vaccine eligibility reduced dementia risk for men, in contrast to the paper from Wales (if anything, females have the weaker effect here, although the difference by sex is not statistically significant).

Source: Pomirchy M, Bommer C, Pradella F, Michalik F, Peters R, Geldsetzer P. Herpes Zoster Vaccination and Dementia Occurrence. JAMA. Efigure 3, Supplementary Material

Why the different results by sex across studies?

It’s hard to know for sure. If there are true sex differences in the biology of immune responses and dementia, we’d expect this to show up consistently across studies. This highlights the importance of replicating studies in different populations, so we can better tell which results stand the test of time (or not). Sometimes effects will be different across populations for substantive reasons—maybe Australians get more Vitamin D from sunshine than people in Wales, and this interacts with the immune system to change the shingles-dementia link. Or the results could be different just by chance--a few more men than women in the “treatment” group could have gotten an unllikely roll of the dementia dice in Wales, making it look like the vaccine didn’t reduce their dementia risk. But if we ran the same “experiment” many more times, the beneficial effect for men would be there most of the time. With more replications the true pattern should emerge, but for now I’m happy to see some evidence that the protective effect extends to men.

Overall, this study gives us reassuring confirmation of the recent finding that the shingles vaccine reduces the risk of dementia. If the findings from Wales were just a fluke, we wouldn’t expect to see a similar pattern in Australia. Like the previous study, the research design is strong for making a causal inference (although we should always maintain some healthy skepticism). The Australian data isn’t quite as strong on sample size and measurement of the key variables, but it still provides another good test case of this intriguing infection-brain relationship.

BOTTOM LINE:

This study makes us slightly more confident that the link between the shingles vaccine and reduced risk of dementia is real. It would be really good news if simple intervention (that has other benefits by preventing shingles) also reduces dementia risk.

Science progresses not by leaps but by persistent testing of our hypotheses (and trying to prove them wrong). It’s great to see this replication, and I hope it leads to more direct testing of the effects of vaccines on cognition and dementia through clinical trials and other interventions.

Stay well!

Jenn

---

In case you missed it: