Flu vaccination linked to 40% reduced risk of Alzheimer’s disease

> lets consider a bunch of real life factors. Wealth, Age, Sex, Race, Educational Attainment, Insurance. Just the basic stuff anyone would think about

The study measured an extensive list of covariates [1]. The data source, "medical claims, pharmacy claims, administrative claims, and laboratory result data for privately insured or Medicare Advantage with Part D enrollees who have both medical and prescription-drug coverage" [2], excludes the uninsured and those without Medicare Part D, so that's not a confounder.

I'm having trouble finding a bias that would create this effect size other than reverse causation, i.e. people developing AD forget to get their flu shots.

[1] https://content.iospress.com/download/journal-of-alzheimers-...

[2] https://content.iospress.com/articles/journal-of-alzheimers-...

This is a published journal article that took those covariates into account. Off-hand comments dismissing articles with a correlation vs. causation quip is so expected and tiresome, especially since you can make these comments without reading the actual article.

Epidemiologist in training here... There are quite a few comments in this thread already jumping on the 'correlation != causation' train. While that is true, I'd like to clarify a couple things:

1. The journal article didn't suggest it was causal. But such a correlation with such a large population warrants publication and further research into causation.

2. literally the first thing that any epidemiologist would consider is potential confounders. There is a big list of covariates they included into their model here: https://content.iospress.com/articles/journal-of-alzheimers-...

There are quite a few things that can be done to alleviate potential false correlations: DAGs, prior literature, removing confounders, and including covariates are all things at disposal.

3. Such a large sample size + previously reported findings + an inclusion of enough covariates still doesn't == causation, BUT it's important to publish and shout about so we can then look into the potential biological underpinnings that may cause this. Which by the way, those experiments may still use data science techniques.

4. If you are actually interested, there is a whole topic of this called 'causal inference' with one famous criteria list called the 'Bradford Hill Criteria': https://en.wikipedia.org/wiki/Bradford_Hill_criteria. This list is often argued about.

5. If all of this information was new to you, please stop spouting 'correlation != causation'. You probably don't know as much as you think

That's indeed, a very impressive list of covariates factored in here.

Hmmm... figuring out a possible mistake would require quite a bit of creativity. Creativity that's beyond my abilities at least, lol.

Note that having such a list hampers a lot of "creative" attempts to take down the study. For example, Tobacco Use is very clearly related to the Republican/Democrat divide, and is therefore an adequate stand-in for the effects of political views (in case political views have an effect on flu-vaccinations / Alzheimer's).

So anything "correlated" to these other covariates (ex: Red vs Blue states vs Tobacco) is "already factored in". As such, having a very, very big list like this helps dramatically.

A lot of armchair scientists in the comments. It's quite striking effect but why is it published on IOS press? Looks like there might be more worth exploring in that dataset. What are the ways in which their result could be an artifact?