A recently published preprint study shows that the shingles vaccine clearly reduced hospital-coded shingles but did not reduce hospital-coded dementia over up to eight years. As can be expected, anti-vaxxers have seized on this preprint as evidence that the shingles vaccine does not affect the risk of dementia.

Of course, a large body of evidence contradicts this non-peer-reviewed preprint, but that doesn’t stop people who want to find anything that refutes the effectiveness of vaccines.

I’ll review the study and present its strengths and weaknesses for those following research on the effect of the shingles vaccine on dementia risk.

Study design

The design is strong for causal inference by observational standards because it exploits a fixed date-of-birth eligibility threshold, comparing people born just before versus just after September 1933 under the same policy.

The sample is unusually large: the primary English Hospital Episodes Statistics (HES) analysis included 6,340,214 individuals born within eight years of the cutoff, about 18 times as large as the earlier Welsh cohort the study sought to replicate.

The analysis estimated an intention-to-treat effect of eligibility, not the effect of actually receiving the vaccine, because HES lacked individual vaccination records.

Main findings

For dementia, the primary estimate was essentially null at -0.06 percentage points over eight years. That confidence interval argues against any large protective effect in this dataset, spanning only about a 3% relative decrease to a 2% relative increase around the observed 13.6% hospital-coded dementia risk.

The null result held across alternative dementia code definitions, alternative denominators, exclusion of the boundary month, fixed bandwidths, alternative RD estimators, and placebo cutoffs.

The one apparent positive result was a 1-year signal, but the authors make a persuasive case that it is not biologically credible because hospital dementia coding lags diagnosis, vaccination rollout was gradual, and a placebo cutoff also showed a nominal signal.

Strengths and limitations of the study

The study uses a quasi-experimental design and data around the policy cutoff for recommending the shingles vaccine. This has fewer confounders than standard association studies. It also has a large sample size, which yields higher precision than smaller studies.

But this study has a major limitation. The HES dataset lacks primary care information (where most dementia diagnoses are made), prescription data, and complete death data. In other words, the study lacked information on most dementia diagnoses. Other studies, which I have discussed in previous articles, captured most dementia cases, which gives a more accurate picture of the change in dementia risk in vaccinated and unvaccinated groups.

The most serious limitation is that hospital-coded dementia is an imperfect proxy for incident dementia, and the dataset cannot identify vaccine uptake or estimate complier effects.

Another important point is that this study examined the live virus vaccine, which is no longer available. Most studies I discussed in other articles examined the current recombinant vaccine.

Summary

Overall, this could be a careful and fairly convincing negative replication: the causal design is strong, the positive control works, and many sensitivity analyses yield null results. However, hospital-based dementia ascertainment and missing vaccination records still limit the inference. I think this is a critical limitation compared with other studies that have shown a strong link between the recombinant vaccine and a lower risk of dementia.

Anti-vaccine activists love to focus on one study to make a conclusion. As a scientist, I assess the study’s quality and add it to the body of evidence indicating that the shingles vaccine reduces the risk of dementia.

Because this study only examined hospital-based dementia diagnoses, it clearly ignores data from primary care. Based on this alone, I am not sure that this study offers much information about the shingles vaccine and dementia.

Furthermore, this study has not been peer-reviewed, so other scientists will have a chance to critically evaluate the validity of the data.

We should also not forget that the shingles vaccine prevents shingles, a dangerous, painful, and disfiguring disease. So, its effect on dementia is a huge bonus, given the substantial body of data contradicting this unpublished study.

Help the Skeptical Raptor keep writing about vaccines

There are four ways you can help contribute to keeping this website running at full speed:

1. GoFundMe.
2. Subscribing via Patreon.
3. Buying your favorite stuff on Amazon.
4. Sharing this article on Social Media.