I took my time reading and writing about the Swedish Tylenol-autism study because I wanted to make sure that I wasn’t just rubber-stamping some results I liked.

And now I can confidently say: this paper is very well-done and very well-written. Here's my review. This study is clearly and carefully documented, uses a very large and rich dataset, and applies a wide range of sensitivity analyses to check for potential biases.

Drinking raw milk is a really bad idea. Only 3% of Americans drink raw milk, but they experience 96% of all milk-related infections.

https://twitter.com/beyoncegarden/status/1940167800938799248

You are *840 times* more likely to get sick from drinking raw milk as you are from drinking pasteurized milk.

Having read the MAHA report, what stands out most to me as an epidemiologist is how inexplicably sloppy it is.

The US government collects data on every health condition you can think of, and the report doesn’t use any of it. At one point, they want to talk about increasing rates of childhood chronic disease. The government collects this data annually. It is publicly available.

For some mindboggling reason, they chose not to use it.

RFK Jr is mandating vaccines get tested against placebo as if this is some bold new idea.

The truth is we do test vaccines against placebo.

But when you have a working vaccine, it is seriously ethically fucked up to continue to withhold it from kids who might otherwise die. This means if a vaccine vs placebo trial is successful, all the placebo arm participants will be informed they got placebo and offered the vaccine.

Because withholding the vaccine from them just so you could do long-term vaccine vs placebo follow-up would be seriously fucked.

FACT CHECK: This viral graph claims to *prove* vaccines cause autism.

But there's something fishy about the data! Let's dive in. The CDC only started collecting data on autism in 2000.

This is what those data👇 look like. A much smaller increase!

IMO, there is very little that’s more dangerous than tyrants acknowledging obvious problems everyone can see exists, because “reasonable” people then assume the tyrants will fix these problems.

The tyrants will not. They are always and only going to advance their own interests. For example: Tyrants do not like universities because universities are an alternative center of free speech and thought and because education has the power to overturn tyrants.

I did a couple raw milk posts this week, and a frequent comment was: “if babies can drink raw milk, why can’t adults?”

It shouldn’t need to be said but:

Human mothers are NOT cows. Human milk is NOT the same as cow milk.

NEVER give a human baby “raw” cow milk. Also, human babies CAN get sick from human milk but it’s also their best food source—even though it is sometimes not an option for all babies or parents.

So, let’s talk about how human babies can get sick from human milk.

I love twitter bc where else can I, an person with two graduate degrees from Harvard in both infectious disease epidemiology and biostatistics, be assured I am wrong about infectious disease epi and statistics by a software engineer who has “read multiple FDA package inserts” sorry if that’s elitist, but I worked really really really hard to know more about this stuff than the average person, so 🤷🏼‍♀️

Niche nerdy tweet incoming:

I’m not at all sure about this instrument.

The “no defiers” assumption seems unlikely to hold — is there really no possible couple who would have divorced if the husband’s workplace stayed the same but not if it had hired more women?

https://twitter.com/jpubecon/status/1840460482240553314

Also, the “treatment” is “parental divorce” which sounds like a single “point” exposure but it’s not. It’s time-varying.

The workplace gender data is daily (wow!) so every day is a new opportunity for the gender-ratio to affect divorce — i.e. timing of divorce likely violates the exclusion restriction assumption.

PUBLIC HEALTH MINUTE:

Honey is sweet and sticky and wonderful and everyone loves it, and babies are sweet and sticky and wonderful and everyone loves them.

But you should NEVER ever give honey to babies.

Why?

The short answer is: botulism.

The longer answer: a 🧵 What is botulism?

Botulism is the general name we give to several different diseases caused by the bacteria Clostridium botulinum.

This includes foodborne botulism, wound botulism, ‘iatrogenic’ (aka cosmetic) botulism and *infant* botulism.

If everyone you know is suddenly getting “food poisoning”, just FYI that it is probably not actually food poisoning but is instead norovirus.

Norovirus (sometimes called “stomach flu” but not actually flu) is seasonal, and right now in the US it is norovirus season. Norovirus symptoms are mostly vomiting, diarrhea, and cramps. But some people also have headache and fever.

Typically, symptoms last only a couple days, but for very young kids, very old people, or people who have weakened immune systems for any reason can be sick for longer.

I am a methodologist. Methods is a speciality. We make recipes for others to use.

When I publish methods, I include relevant code, equations, assumptions, & interpretations. I *intend* for people to re-use these.

Citation is expected. Quotation marks would be ridiculous. For eg, if you are estimating a per-protocol effect of treatment adherence with treatment-confounder feedback using inverse probability weighting, I’ve shared my code & my papers with correct language for your methods & results sections.

Use it, cite it. Thats why it’s there.

Me, an epidemiologist: research shows widespread mask use reduces respiratory infections

Them, a rando: OMG WHY DO YOU WANT EVERYONE TO ALWAYS MASK EVERYWHERE FOREVER?!!!! Me, an epidemiologist: research shows widespread mask use reduces respiratory infections

Them, another rando: oh yeah? Well one time I saw a mask and then I got sick, so how do you explain that, huh?!

When people talk about masks & mask mandates for preventing respiratory infections, they love to talk about improper mask wearing.

And it’s true not everyone wears good masks correctly. But it’s also true that MOST people wear good masks well.

How many is most? About 85%. Data from *Kentucky* show that during their mask mandates, 95% of people in indoor public spaces were seen to be wearing any mask.

And 87% of people were wearing masks that fully covered mouth and nose.

journals.plos.org/plosone/articl…

Out now in Osteoarthritis and Cartilage: our new paper presents a graphical causal model for treatment of pain in knee osteoarthritis.

Open access link: oarsijournal.com/article/S1063-…
We outline a 6-step process for creating causal graphical models using scientific evidence and expert knowledge.

Identify core variables, create a list of potential covariates, conduct a semi-systematic literature review, discuss with experts, synthesize, and visualize.

Nate Silver’s blog👇🏼 is a nice example of using triangulation for causal attribution.

He lays out starting assumptions, then varies his assumptions to see what changes.

He’s not estimating a causal effect. Instead, he’s using descriptive stats to hypothesize about the cause.

https://twitter.com/natesilver538/status/1708952716502786065

Despite this type of triangulation being a key and crucial part of science (and just like, life!), we often don’t spend enough time teaching students how to do it in a reasoned and systematic way.

THINK LIKE AN EPIDEMIOLOGIST:
Science is not magic. It always requires assumptions.

When done perfectly, some types of studies make fewer assumptions. But no one’s perfect.

We can’t decide quality from study type alone: a good observational study can be better than a bad RCT. @benedictecallan I like to call this the “mad scientist target trial framework”.

Not “what RCT would you do”, as a starting point for observational study design.

What experiment would the mad scientist version of you, as a start for designing any study, experiment, observation, or model.

https://twitter.com/epiellie/status/1704114763171565744

Our @NEJM study showed school mask mandates *reduce* COVID illness & absences.

Others disagreed. But we take scientific critique seriously. So we checked everything suggested to us. And now we’ve published an updated Appendix with the details.

Our results hold: Mandates work. Here’s the link to our original study:

nejm.org/doi/full/10.10…

SUNDAY PUBLIC HEALTH MINUTE:

Ever heard “it takes a village to raise a child”?

Well, it also takes a village to start an outbreak. And to stop one.

How do we teach our communities better public health? The answer: public health literacy.

A 🧵 What is “public health literacy”?

Despite the fact that ensuring good public health is one of the most important functions of society, it turns out no one* has ever really defined “public health literacy”!

___
*We’ve found <5 scientific articles ever on public health literacy!

Last month, RFK Jr suggested COVID was an “ethnically targeted bioweapon”.

His claims were racist and wrong. But they also weren’t unique.

Race-based theories of COVID risk have spread almost as easily as the virus itself.

A thread 1/🧵 As we write in our newest article in Scientific American:

“Fear and bigotry make it surprisingly easy to convince people of racial and ethnic differences in disease susceptibility”

https://t.co/P0QTi6R4uCscientificamerican.com/article/racist…

SUNDAY PUBLIC HEALTH MINUTE: Ever seen the phrase “circulating vaccine-derived polio” and wondered what it means?

Time for some real talk about polio and the polio vaccines.

Grab your coffee, snack, and a cozy place to sit — this is gonna be long!

https://twitter.com/epiellie/status/1690349550601371649

We’ll get to the details, but the bottom line is:

Circulating vaccine-derived polio is polio infection that happens because of the vaccine — but it happens mostly to unvaccinated kids*

_____
*or under-vaccinated or immunocompromised kids