sharing a letter i wrote to the NYT about my concerns on the same article. they did not publish it, so no copyright issues.
I read with interest Pam Belluck’s article entitled Shingles Vaccine can Decrease Dementia Risk, Study Finds in the April 3rd paper. To make a causal statement that one action leads to an outcome, one must conduct a randomized trial. However, the scientific study from Wales, descrbied in the article, which links receipt of the shingles vaccine to dementia risk is an observational study, not a randomized one. Observational studies have many well-known limitations, and at best can find associations, but—importantly—cannot determine cause and effect. A well known example of this potential pitfall is the former popularity of estrogen therapy for reducing cardiac risk, which in the 90s was added to the cardiology guidelines based on multiple observational studies. However, when randomized clinical trials were finally done, we learned that there was no benefit of estrogen on cardiovascular risk. The association was likely from the “healthy user” effect, people that seek medical care and therapies, such as shingles vaccine or estrogen therapy, are also likely to be vigilant about other aspects of their health. For example, they are more likely to eat a healthy diet, and get regular physical activity, which are known to reduce dementia risk. Thus, while this study is interesting, it does not support the inference, stated in the headline, that shingles (or any other) vaccine can reduce dementia risk.
79 really is very near end of life so I would not expect it to tell us a lot. Men are typically less frail late in life even if it is shorter but satistically seem to get dimentia less often even correcting for the shorter life.
I have read that rates of dimentia seem to be reducing without an explination. Possibly better treatment for heart disease or maybe the widespread of the shingles vaccines or the trend stops or reverses and we never know.
I do plan to get mine next year at 50 because shingles sounds quite unpleasant.
I suspect the Shingrix vaccine could prevent dementia if it prevents a shingles outbreak in the cranial nerves. My episode of shingles in 2004 affected my face and I lost motor control of my eyes. It also turned me into a mental retard for about 3 months. Fortunately I recovered. I took the Zostavax when it became available, and then the Shingrix when it became available. I suppose those whose shingles is limited to their torso would not be at added risk for dementia because of it.
Very interesting article, thank you. Like you I heard about this research from the non scientific media, and was interested in looking at the paper. I haven’t read it yet, but wanted to comment on the “discontinuity” graph you showed. I have spent a lot of my career evaluating statistics. Looking at the graph, my impression is that the entire effect hangs on one point near the 1933 threshold. It seems that except for that point, the data is better fit with a single line. It’s fairly evident by eye.
Call me skeptical. I wouldn’t hang a big result on a single outlier.
How was this a natural experiment? I think this was nationalized health care playing God. Why would it be unethical to run an RCT on informed, willing subjects yet ethical and "natural" to arbitrarily forbid people from getting the vaccine?
Sadly, journalism needs to exaggerate and sensationalise in order to achieve max engagement. And sadly this does a huge disservice to the sharing of actual insight and wisdom.
Thank you for this post Dr. Cifu. According to what I have read, the NNT for Shingrix for reducing the risk of shingles is ~32. However, often the larger concern is the potential for developing postherpetic neuralgia. Shingrix has an NNT of ~ 333 in 3 yrs for that. Or in other words "for every 333 vaccinated with SHINGRIX, 10 shingle cases (age ≥50 years) and 1 PHN cases (age ≥50 years) were prevented over ~ 3 yrs." Do you find that number compelling? Do patients? When I tell them this, they are not as enthusiastic to get vaccinated. Regards, Cristi BSc (Pharm), RPh
Ealier this year I published an article demonstrating evidence that superinfection and post-translational protein modifications can account for transplant rejection. I spent 34 years in this field and concluded that exposure to mismatched organs did not account for rejection. Reproducible evidence of viral superinfection, which can include components from vaccines-particularly influenza- can account for every instance of organ rejection (https://doi.org/10.1016/j.trim.2025.127197). More recently, I've used the same algorithm to provide evidence for viral superinfection accounting for autoimmune diseases including Alzheimer's (unpublished). Varicella happens to be one of the culprits. Since I was fired for pursuing this work, I can only provide in silico data at this point. Suffice to say that the story on vaccine safety and efficacy is not over just yet. As long as work like this is censored from scientific investigation and public knowledge, it will always just be conspiracy theory... when following the science suggests otherwise.
Newer theories suggest all / most autoimmunity is secondary to an unresolved infection - parasite ,e.g. Lyme, or other. There is still way too much we don't understand about patho / physiology...
Never mind that evaluating patients / studies based only on age, demographics, etc., and not on nutrient status, mobility, strength, etc .... How can we possibly know what myriad of factors predispose someone to a disease / more severe disease state?
And... Why must we keep trying to throw the head of HHS under the bus?
I only had one patient with G-B during my 40 years of medical practice. I had at least a dozen patients either die or be grievously injured in automobile accidents. My advice would be never to drive or be driven in an automobile.
The paper itself is a turgid mass of self-congratulatory twaddle that cloaks its critical methodological weaknesses in a blizzard of camouflaging analysis. And that's being kind.
The authors simply assume their primary outcome, they elide what they "proved" with what they wanted to prove, and they flatly assume causality with the time-honored technique of simply saying that they excluded all other possibilities.
I'll describe the basic methodology without the camouflage (it really deserves its own fisking, which I may take a shot at at some point, but time is short). The authors took a large observational data set from Wales and found a "regression discontinuity" in diagnoses of dementia - a line that they thought should be smooth but had a step point that they correlated to a specific date of birth. That date of birth (in 1933) was significant because in 2013 the Welsh public health authorities made Zostavax (the not-as-well-regarded predecessor to Shingrix) available to individuals born after that date, but not before. Vaccine uptake was brisk because, frankly, shingles sucks. Zostavax actually kind of sucks too, but that's neither here nor there.
The regression discontinuity was only apparent for women, despite the fact that both sexes are susceptible to shingles and there's no obvious reason why there should be a difference. This alone should cause skepticism and deeper analysis, but the authors simply note it and move on.
They then commit their cardinal error: they assume their desired endpoint. Remember, they want to show that shingles vaccination reduces dementia risk. But what they actually showed was that simply having the vaccine available reduced the dementia risk in women. They didn't look to see whether the risk tracked actually receiving the vaccine. Instead, they simply divided by the overall population uptake of the vaccine and increased the effect commensurately.
The critical section is as follows: "Using our regression discontinuity approach, we estimate that the effect of being eligible for the zoster vaccine is a 1.3 (95% CI = 0.2–2.7; P = 0.022) percentage points absolute and 8.5% (95% CI = 1.9–15.1) relative reduction in the probability of a new dementia diagnosis over 7 years (Fig. 3a). Scaled to account for the fact that not all those who were eligible received the vaccine, we find that actually receiving the zoster vaccine reduced the probability of a new dementia diagnosis by 3.5 (95% CI = 0.6–7.1; P = 0.019) percentage points, corresponding to a relative reduction of 20.0% (95% CI = 6.5–33.4) (Fig. 3b).".
But this assumes that the vaccine caused the effect! By assuming that the risk reduction concentrated in the almost 50% of the population who received the vaccine, and assuming that if everyone had gotten the vaccine they would have had a similar result, they inflate a modest result into a large one. Imagine if only 20% of the patients had received the vaccine: in that case, the effect size would have quintupled! And if nobody at all had received the vaccine, the multiplier would have been infinite, and I assume dementia itself would have ceased to exist.
This is blatant mathematical illiteracy in itself and alone would cause the paper to be utterly invalid. The rest of the paper and the discussion around it is self-serving window dressing and outlandish claims built on a foundation of sand.
The fact that the New York Times didn't see this or any of the other problems in the paper (and there are more than a few) isn't surprising. But it was, at least, an interesting exercise in "this seems wrong - can I find out why?" So I thank both Vinay and Adam for that.
sharing a letter i wrote to the NYT about my concerns on the same article. they did not publish it, so no copyright issues.
I read with interest Pam Belluck’s article entitled Shingles Vaccine can Decrease Dementia Risk, Study Finds in the April 3rd paper. To make a causal statement that one action leads to an outcome, one must conduct a randomized trial. However, the scientific study from Wales, descrbied in the article, which links receipt of the shingles vaccine to dementia risk is an observational study, not a randomized one. Observational studies have many well-known limitations, and at best can find associations, but—importantly—cannot determine cause and effect. A well known example of this potential pitfall is the former popularity of estrogen therapy for reducing cardiac risk, which in the 90s was added to the cardiology guidelines based on multiple observational studies. However, when randomized clinical trials were finally done, we learned that there was no benefit of estrogen on cardiovascular risk. The association was likely from the “healthy user” effect, people that seek medical care and therapies, such as shingles vaccine or estrogen therapy, are also likely to be vigilant about other aspects of their health. For example, they are more likely to eat a healthy diet, and get regular physical activity, which are known to reduce dementia risk. Thus, while this study is interesting, it does not support the inference, stated in the headline, that shingles (or any other) vaccine can reduce dementia risk.
79 really is very near end of life so I would not expect it to tell us a lot. Men are typically less frail late in life even if it is shorter but satistically seem to get dimentia less often even correcting for the shorter life.
I have read that rates of dimentia seem to be reducing without an explination. Possibly better treatment for heart disease or maybe the widespread of the shingles vaccines or the trend stops or reverses and we never know.
I do plan to get mine next year at 50 because shingles sounds quite unpleasant.
I suspect the Shingrix vaccine could prevent dementia if it prevents a shingles outbreak in the cranial nerves. My episode of shingles in 2004 affected my face and I lost motor control of my eyes. It also turned me into a mental retard for about 3 months. Fortunately I recovered. I took the Zostavax when it became available, and then the Shingrix when it became available. I suppose those whose shingles is limited to their torso would not be at added risk for dementia because of it.
Does association prove causation? Churnalism at its best.
moreover, the vaccine in question was not Shingrix!
you might say that is not important but there are in fact distinct differences
Very interesting article, thank you. Like you I heard about this research from the non scientific media, and was interested in looking at the paper. I haven’t read it yet, but wanted to comment on the “discontinuity” graph you showed. I have spent a lot of my career evaluating statistics. Looking at the graph, my impression is that the entire effect hangs on one point near the 1933 threshold. It seems that except for that point, the data is better fit with a single line. It’s fairly evident by eye.
Call me skeptical. I wouldn’t hang a big result on a single outlier.
Google "heron, superinfection, HLA"
How was this a natural experiment? I think this was nationalized health care playing God. Why would it be unethical to run an RCT on informed, willing subjects yet ethical and "natural" to arbitrarily forbid people from getting the vaccine?
Sadly, journalism needs to exaggerate and sensationalise in order to achieve max engagement. And sadly this does a huge disservice to the sharing of actual insight and wisdom.
This was one of your better articles until you had to throw a gratuitous insult at RFK Jr.
💯
Gratuitous yes, but also kind of subtle.
I had to re-read the article to find it! Nice sleight of hand. Touché!
Thank you for this post Dr. Cifu. According to what I have read, the NNT for Shingrix for reducing the risk of shingles is ~32. However, often the larger concern is the potential for developing postherpetic neuralgia. Shingrix has an NNT of ~ 333 in 3 yrs for that. Or in other words "for every 333 vaccinated with SHINGRIX, 10 shingle cases (age ≥50 years) and 1 PHN cases (age ≥50 years) were prevented over ~ 3 yrs." Do you find that number compelling? Do patients? When I tell them this, they are not as enthusiastic to get vaccinated. Regards, Cristi BSc (Pharm), RPh
Reference: https://www.rxfiles.ca/RxFiles/uploads/documents/Shingrix_QandA.pdf
This is a great piece, thank you
Ealier this year I published an article demonstrating evidence that superinfection and post-translational protein modifications can account for transplant rejection. I spent 34 years in this field and concluded that exposure to mismatched organs did not account for rejection. Reproducible evidence of viral superinfection, which can include components from vaccines-particularly influenza- can account for every instance of organ rejection (https://doi.org/10.1016/j.trim.2025.127197). More recently, I've used the same algorithm to provide evidence for viral superinfection accounting for autoimmune diseases including Alzheimer's (unpublished). Varicella happens to be one of the culprits. Since I was fired for pursuing this work, I can only provide in silico data at this point. Suffice to say that the story on vaccine safety and efficacy is not over just yet. As long as work like this is censored from scientific investigation and public knowledge, it will always just be conspiracy theory... when following the science suggests otherwise.
DOI is not found. I would be happy to read.
Sorry, https://doi.org/10.1016/j.trim.2025.102197.
Newer theories suggest all / most autoimmunity is secondary to an unresolved infection - parasite ,e.g. Lyme, or other. There is still way too much we don't understand about patho / physiology...
Never mind that evaluating patients / studies based only on age, demographics, etc., and not on nutrient status, mobility, strength, etc .... How can we possibly know what myriad of factors predispose someone to a disease / more severe disease state?
And... Why must we keep trying to throw the head of HHS under the bus?
In case I wasn't clear, I'm supportive of the secretary.
Ah, yes. My apologies. I wrote in haste and meant to direct the last comment to Dr. Cifu.
I only had one patient with G-B during my 40 years of medical practice. I had at least a dozen patients either die or be grievously injured in automobile accidents. My advice would be never to drive or be driven in an automobile.
The paper itself is a turgid mass of self-congratulatory twaddle that cloaks its critical methodological weaknesses in a blizzard of camouflaging analysis. And that's being kind.
The authors simply assume their primary outcome, they elide what they "proved" with what they wanted to prove, and they flatly assume causality with the time-honored technique of simply saying that they excluded all other possibilities.
I'll describe the basic methodology without the camouflage (it really deserves its own fisking, which I may take a shot at at some point, but time is short). The authors took a large observational data set from Wales and found a "regression discontinuity" in diagnoses of dementia - a line that they thought should be smooth but had a step point that they correlated to a specific date of birth. That date of birth (in 1933) was significant because in 2013 the Welsh public health authorities made Zostavax (the not-as-well-regarded predecessor to Shingrix) available to individuals born after that date, but not before. Vaccine uptake was brisk because, frankly, shingles sucks. Zostavax actually kind of sucks too, but that's neither here nor there.
The regression discontinuity was only apparent for women, despite the fact that both sexes are susceptible to shingles and there's no obvious reason why there should be a difference. This alone should cause skepticism and deeper analysis, but the authors simply note it and move on.
They then commit their cardinal error: they assume their desired endpoint. Remember, they want to show that shingles vaccination reduces dementia risk. But what they actually showed was that simply having the vaccine available reduced the dementia risk in women. They didn't look to see whether the risk tracked actually receiving the vaccine. Instead, they simply divided by the overall population uptake of the vaccine and increased the effect commensurately.
The critical section is as follows: "Using our regression discontinuity approach, we estimate that the effect of being eligible for the zoster vaccine is a 1.3 (95% CI = 0.2–2.7; P = 0.022) percentage points absolute and 8.5% (95% CI = 1.9–15.1) relative reduction in the probability of a new dementia diagnosis over 7 years (Fig. 3a). Scaled to account for the fact that not all those who were eligible received the vaccine, we find that actually receiving the zoster vaccine reduced the probability of a new dementia diagnosis by 3.5 (95% CI = 0.6–7.1; P = 0.019) percentage points, corresponding to a relative reduction of 20.0% (95% CI = 6.5–33.4) (Fig. 3b).".
But this assumes that the vaccine caused the effect! By assuming that the risk reduction concentrated in the almost 50% of the population who received the vaccine, and assuming that if everyone had gotten the vaccine they would have had a similar result, they inflate a modest result into a large one. Imagine if only 20% of the patients had received the vaccine: in that case, the effect size would have quintupled! And if nobody at all had received the vaccine, the multiplier would have been infinite, and I assume dementia itself would have ceased to exist.
This is blatant mathematical illiteracy in itself and alone would cause the paper to be utterly invalid. The rest of the paper and the discussion around it is self-serving window dressing and outlandish claims built on a foundation of sand.
The fact that the New York Times didn't see this or any of the other problems in the paper (and there are more than a few) isn't surprising. But it was, at least, an interesting exercise in "this seems wrong - can I find out why?" So I thank both Vinay and Adam for that.
The only thing I understood from your comment was "shingles suck." And your name is fantastic Dread Overlord🤣!
"turgid mass of self-congratulatory twaddle" Love it.
I should have just written that and called it a day!
Hi! I should get a shingles vaccine, but what about the black box warning it has regarding Guillian-Barre? Is it worth the risk?