Pascal Geldsetzer, MD, PhD, Assistant Professor of Medicine in Primary Care and Population Health at Stanford, joins The Future of Medicine to discuss a striking line of research: evidence that shingles vaccination may reduce the risk of dementia.
Geldsetzer and his team study “natural experiments” — real-world situations that can allow researchers to get closer to determining cause and effect without conducting a traditional randomized clinical trial. In the case of shingles vaccination, age-based eligibility rules created unusually comparable groups of people: those born just before a cutoff were not eligible for the vaccine, while those born just after it were.
Using these natural experiments, Geldsetzer’s team has found evidence across several countries that shingles vaccination is associated with a substantial reduction in dementia risk. In one analysis, the researchers estimated that vaccination reduced the risk of dementia over seven years by about 20%. Similar patterns have appeared across different datasets and health systems, as well as in studies examining mild cognitive impairment and outcomes among people already living with dementia.
The conversation explores why a vaccine designed to prevent shingles might affect the brain. Geldsetzer discusses possible roles for the virus that causes chickenpox and shingles, inflammation, blood vessel health, and broader effects of vaccination on the immune system. He also explains an important open question: much of the strongest evidence comes from an older live-attenuated shingles vaccine, so additional research is needed to understand whether newer shingles vaccines have similar effects on dementia.
The Intro
“We estimate that shingles vaccination reduces your risk of dementia over 7 years by 20%. This is a simple, one-off, inexpensive, readily scalable intervention”
Dr. Pascal Geldsetzer was named one of Time magazine's 100 most influential people in health and medicine.
“We see benefits also among those who already have dementia. We see reductions in your probability of dying from dementia.”
His research suggests that a shingles vaccine, a shot you can get at any pharmacy, may lower your risk of developing dementia by 20%.
“And it's just an effect that we see in data set after data set in different countries, different health system settings, for different dementia-related outcomes. It's not a fluke finding. It keeps replicating.”
We explore how he uncovered this, why the finding keeps appearing in country after country, and what it means for you and your loved ones.
Welcome to Stanford Department of Medicine's inside look at the future of medicine.
The Transcript
Dr. Errol Ozdalga: Pascal Geldsetzer, thank you so much for being with us today.
Dr. Pascal Geldsetzer: Well, thanks for having me.
Dr. Ozdalga: Tell me, I want to learn more about the type of research you do. In learning about you, I found that you do something called natural experiments. What's a natural experiment?
Dr. Geldsetzer: Yeah, so it's something that I received an NIH Innovator Award for, to develop and apply to clinical research questions in electronic health record data. It's a way of getting at cause and effect in observational data in a manner that's much more plausible and robust than if we do simple comparisons like we often do in epidemiology and observational data. Generally speaking, we'll compare people who go get a certain medication versus those who don't, or people who get a vaccine versus those who don't. There's a fundamental bad-comparison problem there, right? People who decide to get a certain intervention usually differ in their health motivations, health behaviors, etc. To be able to adjust for all of these potential confounders, we would need perfect measurements of all of these things in our data sources, and that's just not a plausible assumption in most cases. So we're left with associations that we don't know whether they actually reflect cause and effect or just a correlation. That's why people often think of observational data as evidence that's interesting and hypothesis-generating, but what we really need to show cause and effect is a clinical trial where we randomize people. In natural experiments, we try to find situations in nature, in clinical care settings, for example, where there has plausibly been some quasi-randomization. Of course, these can vary in their level of plausibility in a wide variety of ways. I think in the case of the shingles vaccine, we have a really beautiful natural experiment, a textbook case, because we know for sure that the reason you either got the shingles vaccine or not was simply because you were born just a few days earlier or a few days later. That's the basic intuition: these must be good comparison groups. If I take a thousand people born one week and a thousand people born a week later, why should there be anything different about them except for this tiny difference in age? And in the case of the date-of-birth cutoff for shingles vaccination, there's a huge difference in your probability of getting the vaccine.
Dr. Ozdalga: So, if I understand this correctly, in a randomized controlled trial, which is what we typically think of as the gold standard for studying two groups, we artificially create those two groups. We randomly put people in two groups, by throwing a coin, and try to keep them similar. But what you find are situations that naturally already create two groups, so you don't have to worry about all these things being perfect. It's like natural spring water versus filtered water: you're finding the natural spring water and studying that. Does that make sense?
Dr. Geldsetzer: Yes, essentially. We're trying to find situations where this coin toss that we have in a clinical trial, to assign people to the control group or the intervention group, somehow plausibly happened in the health care setting. In the case of shingles vaccination, it was essentially by random chance whether you happened to be born a few days earlier or a few days later, and that is similar to the coin toss where we assign you to a control group or an intervention group in a clinical trial where we actively create good comparison groups.
Dr. Ozdalga: And how do you find a natural experiment? How did you find the ones you've found, and how did you get into this?
Dr. Geldsetzer: Regression discontinuity is the statistical approach we use when we have these threshold-based decisions in natural experiments. We were looking for thresholds in vaccination programs that would allow us to get at the cause-and-effect of vaccines on health outcomes, and we came upon the shingles vaccination program. It's a really unique opportunity, I must say, because we have a situation where just being born a few days earlier or a few days later meant your probability of getting the vaccine jumped from 0% to 50%. Such a sudden, crisp contrast is something we rarely have in clinical medicine, and that provides a really powerful setting to apply these methods.
Dr. Ozdalga: So you've talked, and I want to dig into the amazing work you've done on the shingles vaccine and dementia. Along with that research — also, in preparing for this, I went through your CV and your author publications, and I actually got tired of scrolling. You've done so many publications already, but I got a good flavor of all the things you've been involved with. Usually you're the primary author or the senior author on a lot of work — things like prevalence of non-communicable disease risk factors among middle-aged adults in India, disparities in air-pollution-attributed mortality in the United States, a stepped-wedge randomized trial of HIV pre-exposure prophylaxis in Africa, smoking, blood pressure, and anemia among men in other countries, cervical cancer screening. You've been involved in research on all sorts of things, in all sorts of diseases, in all sorts of parts of the world. I'm curious: of all these things you've studied, what has shocked you the most? What has made you say, "Oh my god"? Was it the shingles vaccine?
Dr. Geldsetzer: Yeah, absolutely. I've explored many different areas of research. I've always been looking for ways of doing research that I think can actually have an impact on populations, right? Just publishing something isn't satisfying to me — I want to create some social good. Ever since we came upon the shingles vaccination program, that's been the entire focus of my group for three years now, and will continue to be moving forward, because I think the impact, the implications for population health, for clinical medicine, for neurodegenerative disease research, would be so enormous that this deserves research attention.
Dr. Ozdalga: That's wonderful. Let's start there. Let's dig into this research you've been doing over the last few years, which has gotten a lot of attention. Tell us — to somebody who hasn't heard about this research, or who doesn't understand the scientific background of it — what was this research you did, and what did it show?
Dr. Geldsetzer: We have been looking at the effect of shingles vaccination on dementia, for prevention of dementia and potentially even treatment of dementia. The reason we're so excited about this research and our findings is not just that we see these large protective effects, but that we're able to use this natural-experiment approach in an extremely unique way to get at causal effect estimates. We think our findings actually reflect cause and effect rather than just correlation, because of this unique situation where people born just a few days earlier or a few days later were either eligible or ineligible for the vaccine. We know, and we can show this in our data, that those are perfect comparison groups. That overcomes the fundamental problem we usually have in medicine, in observational data analyses, in epidemiology — these bad comparisons where we have to compare people who decide to get a certain intervention, a vaccine, or a medication, or where the health care provider decides to provide it. We always have to try to adjust for all sorts of differences and then keep our fingers crossed that we adjusted for enough. We don't have that problem here, and that's why we're so excited about it.
Dr. Ozdalga: And just stepping back a little — I understand you found a cohort in Canada, in Wales, and in Australia, where people were separated by age, and if they were born just days later they would be in the category where the shingles vaccine was offered. You compared that to the group that didn't have the shingles vaccine offered.
Dr. Geldsetzer: Yeah, exactly.
Dr. Ozdalga: Can you tell us — was the risk of dementia significantly lower in those who were in the group that had the shingles vaccine offered?
Dr. Geldsetzer: Yeah, exactly. The beauty of the shingles vaccine is that it was rolled out in such a unique way, not just in one country but in several countries. What's so exciting for us at the moment is that we keep seeing these effects in all these different settings, which gives us much more confidence that what we're looking at here really is a true cause and effect. But yes, essentially — for example, in the UK, if you had your 80th birthday just prior to the start date of the program, you were ineligible, and you remained ineligible for life. You were too old to ever be eligible for the vaccine. But if you had your 80th birthday just after, you were eligible. We have week of birth in our data, so we can see that a week's difference in your age at the time of the vaccination program meant your probability of ever getting the shingles vaccine jumped from 0% to 50%. This results in a huge difference between people who are otherwise perfectly matched — all that's different about them is a week's difference in age.
Dr. Ozdalga: So roughly half of those people who had the opportunity to get the vaccine, got vaccinated. And what was the difference between that group and the group that didn't have the opportunity to take the vaccine?
Dr. Geldsetzer: Right. We estimate that shingles vaccination reduces your risk of dementia over 7 years by 20%. And you have to think about this in the context of it being a simple, one-off, inexpensive, readily scalable intervention. It's not even a medication you have to take every day — not that we have that for dementia — and it's certainly not a diet or physical activity regimen you have to adhere to for decades, which we know people find difficult. This would have huge — and a 20% reduction from such a simple intervention, given the growing dementia burden in the future — the implications are profound. That's why I think we need to invest more into this. We need to pursue this.
Dr. Ozdalga: And you've shown this in many ways, in many countries. You say you continue to see a strong signal. Tell me more about that.
Dr. Geldsetzer: Yeah. Our first paper, published in Nature, used data from Wales. All of the UK rolled out the vaccine this way, and we keep seeing these odds in other UK data sets.
Dr. Ozdalga: There's a graph with a line in the middle where you either meet the cutoff or don't, and it shows you the differences in risk of dementia.
Dr. Geldsetzer: Right. And also the differences in your probability of getting the vaccine — it's a super sharp contrast. This exists across the UK. Then we published a paper in JAMA just a few weeks later, using data from Australia, which is one other country that rolled out the vaccine using this date-of-birth cutoff — of course, a different date-of-birth cutoff. And then, as you were mentioning, we had a paper from Canada in The Lancet Neurology, published just earlier this year. Canada is a really nice setting for this research because they actually used two date-of-birth cutoffs, and only in Ontario. So you can look at both date-of-birth cutoffs to see if you see these effects, and you can compare eligible birth cohorts in Ontario to eligible birth cohorts in other Canadian provinces. If you think there is a protective effect from shingles vaccination, you would expect dementia incidence to start diverging between the eligible birth cohorts in Ontario, where the shingles vaccination program existed, and the same birth cohorts in other Canadian provinces, where there wasn't a shingles vaccination program. And that's what we see in our data.
Dr. Ozdalga: So, to make sure I understand this at a very simple level — there were two different cutoffs where people were given the opportunity to get the vaccine in Ontario. Were there different percentages of people who got it in those groups, or was it just a different time course?
Dr. Geldsetzer: No. At each of these date-of-birth cutoffs, you always see this sudden jump in the probability of ever getting the vaccine, and that jump is what we're exploiting in our natural experiment.
Dr. Ozdalga: So one time course gets the vaccine, you see that jump in probability of getting the vaccine, and then you also see a decrease in the probability of getting dementia if you follow it over years?
Dr. Geldsetzer: Right, exactly.
Dr. Ozdalga: And then you see that same thing happen with the second cohort as well, at the same — So you're seeing the same data being replicated over and over, telling the same story?
Dr. Geldsetzer: Yes. I think that's what's so compelling about this body of evidence we now have on shingles vaccination and dementia — that we see these large effects using this approach, exploiting this really unique setting of shingles vaccination rollouts, which allows us to get at cause and effect in a much more plausible, much more likely way. It's an effect we see in data set after data set, in different countries, in different health system settings, for different dementia-related outcomes. We see it for protection against mild cognitive impairment, for dementia diagnosis in general. We see benefits also among those who already have dementia at the time of becoming eligible for vaccination — we see reductions in the probability of dying from dementia. That's a paper we had in Cell late last year. It's not a fluke finding — it keeps replicating.
Dr. Ozdalga: Speaking of that Cell paper — there's dementia, and then there's pre-dementia, mild cognitive impairment. You also talked about that in that paper. Tell us more about what you saw in the earlier forms of mild dementia.
Dr. Geldsetzer: Yeah. We were looking at individuals without any cognitive impairment, as far as we can tell from the electronic health record. Instead of looking at new dementia diagnoses in the future, we looked at new diagnoses of this pre-dementia, mild cognitive impairment. There, too, we see large reductions in the probability of such a diagnosis, suggesting there are strong protective effects there as well.
Dr. Ozdalga: So there's value in so many different ways in how your brain works, it seems.
Dr. Geldsetzer: Yeah, yeah.
Dr. Ozdalga: I should probably say this — there's a lot of controversy with vaccines in general, but your goal is just to show the data. If the data had said vaccines are dangerous or bad, you'd be talking about that. There's nothing about this — have you gotten any pressure from people simply because they don't trust vaccines? There's been a lot of controversy, especially over COVID, about vaccines.
Dr. Geldsetzer: Well, certainly a bit, but I think overall the reception of this research has been excitement, especially from the Alzheimer's and dementia research community. There's a lot of frustration with the mainstay hypothesis of Alzheimer's disease in that community, and here we would have an intervention based on a different mechanistic hypothesis of dementia — but also one that is safe and has large effects. People are excited about investigating this further, and I think that's really cool.
Dr. Ozdalga: That's been a great question. Can you, on a more simplistic level, tell us what's going on here? I know it's not fully understood why, but there are some thoughts on why this vaccine works for dementia in the ways it does.
Dr. Geldsetzer: Yeah, yeah. So, of course, we know that after you contract chickenpox, the virus remains with you for life. It's in a constant interplay with the immune system, and as it tries to reactivate and is beaten back down by the immune system — and then if it reactivates fully, that's when you get shingles — we know that because this virus hibernates in the nervous system, this causes some inflammation in the nervous system. We also know that shingles can lead to inflammation of blood vessels, and blood vessel health is, of course, an important component of brain health as well. So essentially, this idea that reducing these reactivations through shingles vaccination may have benefits is one type of mechanism we can think of. The other type is that there's a growing body of evidence showing that vaccines seem to have broader effects on the immune system beyond the specific antibody response they've been designed to elicit, and that these broader effects seem to protect you, to some degree, from other infections, and may potentially also have an important role in chronic disease development, including dementia.
Dr. Ozdalga: You mentioned that a lot of these studies used the live attenuated version of the vaccine. In the US we use not the live attenuated version, but there's every reason to believe it would work as well, or even — I think there's some data suggesting it might work even better — at preventing dementia.
Dr. Geldsetzer: It depends on the mechanism we think is at play. If it's about reducing reactivations of the virus, we'd think the newer vaccine would have potentially even larger benefits for dementia than the older vaccine, because it's more effective at reducing these reactivations. But if it's about these broader effects on the immune system, which may well be independent of the virus — and we know these broader effects are often particularly strong for live attenuated vaccines, and the old shingles vaccine is a live attenuated vaccine while the new one is not — then it may be specific to the old, live attenuated vaccine. That's the reason why, for us, the next step needs to be a clinical trial, and ideally we want to use the old shingles vaccine, the live attenuated one, because all the evidence we have from our natural experiments is for this vaccine. Proving that this is really a true causal effect that we see in data set after data set — it would have such huge implications for population health and dementia research. This is something I think we must invest in.
Dr. Ozdalga: Would you, in a clinical trial, include the vaccine we use in the US, to see if it's equally effective to the one that's been studied more?
Dr. Geldsetzer: That would be the ideal trial, of course. The more things you want to test, the larger the trial you need, and the more money you need. It's been a struggle — I'm trying to raise funds from philanthropic foundations to see if we can get a trial of this old, live attenuated, off-patent vaccine off the ground, since there's little commercial interest. I think we really need to turn to foundations and philanthropy to get such a project off the ground. But the more questions you want to test, the more money we need to raise.
Dr. Ozdalga: Would it be reasonable, based on the data we have, to state that the vaccine currently used in the US likely does work similarly to the one that's been studied more?
Dr. Geldsetzer: I don't know — I try to shy away from making official vaccine recommendations in any way. I'm a researcher. I'm trying to generate the evidence, and I think it's up to expert committees that have been specifically put together for this purpose to evaluate whether or not certain recommendations should be issued.
Dr. Ozdalga: So, as an internist, it's not clear to me whether I should tell my patient, "You're going to be protected from shingles," and maybe say there's a possible benefit for similar vaccines in decreasing brain damage, dementia, and mild cognitive impairment, and even in treating dementia. That would be a more reasonable statement until we get more guidance from committees and more information.
Dr. Geldsetzer: I think that's certainly something the evidence supports. Yes.
Dr. Ozdalga: And if it does work the same way for this live attenuated vaccine — a 20% reduction — I mean, this is as strong as, or stronger than, some of the main treatments we use today, like statins or blood pressure medications. We forget that not everybody benefits from a statin — you need to treat a lot of people to see that strong benefit.
Dr. Geldsetzer: And people need to take that medication every day.
Dr. Ozdalga: Right. There are probably more side effects with blood pressure medications, and perhaps statins, than you'd get with a vaccine whose most common side effect is a little pain at the injection site.
Dr. Geldsetzer: Right, right. And you get shingles prevention too.And you get shingles prevention, which is also a nice side effect.
Dr. Ozdalga: Yes, absolutely. Tell me more about, as far as pairing with other interventions goes — I think we're not there yet as far as indications. Maybe down the line, with more research, we may see things change in the US, but right now this vaccine still needs to be focused mainly on shingles, with the possible benefit for dementia. That's kind of where we are right now, talking about the vaccine used in the US. Do you see things changing? It sounds like we'd need more studies specifically on that vaccine to say for certain that it's more than just protection against shingles — it very likely is more, but to say so for sure we need more studies.
Dr. Geldsetzer: Yeah, so unfortunately, for the new vaccine, Shingrix, we haven't had the opportunity for these beautiful, clean natural experiments, because that wasn't the vaccine that was rolled out this way, using date-of-birth cutoffs, and it's only been available for a much shorter time than the older vaccine. So we just don't have that same level of evidence for the newer vaccine. But I think it's a critical research question moving forward — what the effects on dementia incidence are for that newer vaccine.
Dr. Ozdalga: You've done a lot of press work, and this has gotten a lot of attention. Congratulations on being named one of Time's 100 most influential people in health and medicine — that's really amazing. You've told me you've been doing this work for about five years or so. And you already have not just national recognition but global recognition for the work you're doing, and you're still just getting started. That's amazing. So, you've been in the news — what's the hardest thing to communicate about the work you've done so far? Part of the goal is to do the work and generate the information; the other half, which I hope we can help with, is to share that work so it changes more research and changes how we care for our patients. What's the biggest challenge around that second part — the communication?
Dr. Geldsetzer: Yeah. I think it's that people read headlines all the time about, "If I take this vitamin, I have a lower risk of that in the future," or "If I sleep an hour more, I have a lower risk of this," or "If I do this type of exercise versus that type" — people are kind of tired of it, I think, or don't really know what to make of it, and it's a bit overwhelming. What I worry about is that people aren't distinguishing that the evidence we now have for shingles vaccination and dementia, from these types of analyses, is at a very different level than we have for these other headlines, which are usually not based on clinical trials but on epidemiological analyses, where we always have this fundamental bad-comparison problem — that people who decide to take a certain supplement or eat a certain way are different from people who don't, and we can't really control for all of these differences in our analyses. We don't have that problem here, and that's why I think it's important to message to the public that this is different — this is a different level of evidence.
Dr. Ozdalga: Excellent. I know you're working on the vaccine and dementia research we've been talking about, but if you could study any natural experiment in the world — just assume it exists — is there something you've thought, "I would love to investigate this topic"?
Dr. Geldsetzer: To me, wanting to have an effect and a positive impact on population health and on patients — this dementia question has actually been really fascinating, because dementia is something people are really worried about. It's so common, and it's a condition we have so few tools against at the moment. So the promise of a one-off, inexpensive, readily available intervention that could have large effects on such a major future burden — it's actually hard for me to think of what other natural experiment I would love. I mean, I would love more natural experiments to answer all the questions posed by this research, right? There are so many questions we need to investigate — which particular population groups are especially benefiting, what the mechanism is, what the time dynamics are, whether we need booster shots, whether it might apply to some other vaccines given in older age. I'd love beautiful natural experiments for all of these questions. It's a really unique opportunity.
Dr. Ozdalga: What I'm feeling is that you are laser-focused on dementia, which is amazing.
Dr. Geldsetzer: Right, right.
Dr. Ozdalga: Speaking of that — when you did the study, you have this beautiful setup: people either did or didn't get the vaccine, depending on whether it was available to them. I'm assuming you didn't see the same signal for risk of heart attack, or for any other diagnosis — it was uniquely for dementia.
Dr. Geldsetzer: And shingles. And yes, and complications of shingles, like postherpetic neuralgia.
Dr. Ozdalga: Which tells me that signal is unique — it stuck out. Those two signals stuck out among so many other possible signals. Do you look for anything else out there? It sounds like you look for lots of other diagnoses being affected by the vaccine.
Dr. Geldsetzer: Right. We do this to demonstrate that what we're looking at here is likely really true cause and effect. We go through these other common health conditions because, when we have this fundamental bad-comparison problem in epidemiology, we would usually see effects not just on the outcome we're looking at, like dementia, but also on other outcomes — people who self-select into getting vaccinated because they're more health-motivated and have better health behaviors don't just have a lower risk of dementia, they also have a lower risk of heart attacks, etc. Not seeing that in our data, and having these specific effects for shingles and dementia, is really reassuring that what we're looking at here is cause and effect.
Dr. Ozdalga: You mentioned earlier that you've been out of training for five years. You have a medical degree — you went to UCSF, and you got your Master of Public Health and PhD at Harvard. You have a medical degree, and you spend a lot of time doing research. For those who want to follow your path and get into this work, how did you get here, and what advice would you have for early-career trainees wanting to follow the same career path?
Dr. Geldsetzer: I think — I've always wanted to do things that I enjoy and that can have a positive impact on society, and that's sort of how I've made my decisions as I've moved through some of these career decisions. After coming out of medical school, I thought clinical medicine wasn't really something I personally enjoy, that fits my personality. I think research is something where I can have a positive impact, but it's also something I really enjoy doing. And then I made the decision that if I want to have a positive impact, I need to do research well, so I need to get proper training — that's why I went into my PhD program and got a lot of methods and statistics training. Since then, I've just been trying to pursue research topics and research questions that I think can really make a difference.
Dr. Ozdalga: The enthusiasm you have for this work, particularly around shingles and dementia, is clear. Of course, you're going to follow the data wherever it leads, but we all have a good intuition that the data so far is pretty clear, and we'll see what more data shows us. Ten years from now, looking back on how this all started, what would be your hope for what this started? What would you like to see, how would you like this to change how we practice medicine, how we care for people? What's your hope for what this — which kicked off three years ago — leads to over the next ten years?
Dr. Geldsetzer: I mean, my hope is — either way, if we find that somehow this isn't a true cause and effect, I'm open to that, and it's good to know. Of course, it's important to know, so that we don't invest further into this. But of course, my hope is that this could be an intervention to stem the burden of dementia, and that it could be an intervention and a research body that leads to a ton of additional insights that then lead to further therapeutics and preventive tools — if we understand the mechanism better, for example. What could that lead to? I think that — and maybe it's not just about dementia, maybe it's about neurodegenerative disease more generally, about immune aging in older age, aging processes in the nervous system and the immune system — I'd love to see benefits and insights there that can benefit population health and patients.
Dr. Ozdalga: One last question — is there a question you wish you'd be asked, but no one ever does?
Dr. Geldsetzer: That's an interesting question. No, I think — it's been really awesome, I must say. People are excited about this. The media, of course, is excited about it, but I think there's a genuine interest that this could be important and that we need to inform the public about it. So it's actually been really great — it's been a really positive experience, and I've generally been quite impressed with the sort of questions I've gotten.
Dr. Ozdalga: Well, thank you — that makes me feel good. Well, Dr. Pascal Geldsetzer, thank you so much for everything, and thank you for being with us today.
Dr. Geldsetzer: No, thank you for having me — thanks for the opportunity to talk about this.
Dr. Ozdalga: Awesome, thank you.