Media, Substackers, and authorities are screaming bloody murder about the difference between new Covid vaccine guidelines posted by the American Academy of Pediatrics and guidelines from the CDC. For kids under two the AAP recommends the shot. The CDC says not so fast: discuss it with a doctor.

Who’s right? Guidelines are written by panels that review the data, then make judgment calls. The only way to know if you agree is to understand the data. And to do that, we must first revisit a fated moment in Covid history.

For the July 4th weekend in 2021, I was on Cape Cod with my kids. We were supposed to be safe from the virus. In the preceding months more than half the US population had been urgently vaccinated, and in a town full of LGBTQ+ people, artists, and self-styled intellectuals, it seemed almost certain the vaccination rate was way more than half. But the Delta variant hit Provincetown like a tornado, and vaccinated people everywhere got sick, including me. I briefly lost my sense of smell for the first time. But the world lost something else, permanently: the dream of herd immunity.

Here’s why. The MMR vaccine prevents about 97% of measles infections, which is high enough that if 90% or more in a community get the shot, the measles virus has no inroad. While rare ‘breakthrough’ infections may occur, outbreaks are impossible. For the virus, every road is a dead end. That’s herd immunity.

With a prevention rate of 95% in Covid vaccine trials, herd immunity seemed tantalizingly close. Then Provincetown happened. The Cape Cod outbreak signaled to the world that the virus would constantly mutate, and even vaccinated people would routinely be infected. In a single jubilant holiday weekend, the dream of herd immunity came and went like so many fireworks.

But there was some good news: vaccinated people rarely died. Covid vaccination, it seemed, made death or serious illness extremely unlikely. Which is critical for decoding the vaccine’s benefits, and thus the guidelines.

Now, to the data.

The best data in kids comes from a vaccine trial of 4,526 children, a little less than half of whom were between 6 months and two years old. The actual focus of the trial was whether the kids developed Covid antibodies, which they mostly did, but importantly the researchers also reported the vaccine prevented 76% of Covid cases.

Unfortunately, this estimate was based on just 4 illnesses in the vaccine group and 8 in the placebo group, or 1% vs 5%, respectively. That is weak data, and a useful reminder: even in a 2021 trial from the heart of the pandemic, kids get sick with Covid less than adults and experience less severe illness.

But remember what we said. The benefit of the vaccine is not in preventing illness—vaccinated kids will still get sick—it’s prevention of serious illness. Which raises an obvious question: how often do young kids become seriously ill with Covid? In other words, how common is the problem the vaccine aims to prevent?

It's rare. The CDC estimates 1 in every 2000 young children in the U.S. was hospitalized with Covid in 2025. While that is a surprisingly large number (suggesting about 3,500 children hospitalized), it’s also seven times lower than the corresponding number for adults. Moreover, less than half of the hospital stays were for "severe Covid", in which a child received ICU care, oxygen, or other Covid treatments. Indeed, more than half the kids were home in less than two days, suggesting the admission was, at worst, precautionary. But, in theory, vaccination could prevent 76% of those hospitalizations by preventing the initial infection, which would mean 1 in every 2,632 vaccinated kids in the U.S. avoids hospitalization because of the vaccine.

Next we have to quantify the vaccine’s harms, but the data is even sparser. The Pfizer trial of young children reports “severe reactions” in 0.3% (1 in 333), but these were fevers, irritability, and loss of appetite, usually for a few days or less. It takes larger datasets, like the 67,000-person original Pfizer and Moderna trials, to find major harms. The best report on harms from those two trials found roughly 1 in 800 people experienced serious illness due to the vaccines. Unfortunately, current data can’t tell us whether children are more or less harmed than adults, so we’re left to apply adult harm data to kids.

In aggregate, then, a best-case estimate for the vaccine suggests about 0.038% (1/2,632) could benefit by avoiding hospitalization, while 0.125% (1 in 800) would be harmed by a life-threatening or serious illness. Based on these numbers harms are three times more common than benefits.

There you have it! Sort of.

Now take a moment to consider how many judgment calls it took to generate these calculations. To estimate benefits I selected the most flattering possible study, extrapolated tiny numbers to the rest of the world (4 vs 8 kids!), and then accepted a dubious, almost certainly inflated, CDC estimate of hospitalization rates. For harms I ignored all transient ‘reactions’, and was forced to apply adult data to children.

Clearly, there’s room for judgment. It becomes easier to see how two expert panels might differ. If you believe vaccines are under-appreciated and very safe for kids (read: if you’re the AAP), you might conclude the serious harms from the vaccine are likely much less common in kids than in adults. Which could shift the recommendation because even a tiny benefit, or potential benefit, would be bigger and more common than a serious harm.

If you’re a Covid vaccine skeptic, or a pharmaceutical company skeptic, (read: if you’re RFKjr) and you believe Pfizer’s reported benefits and the CDC’s rates of hospitalization are likely exaggerated, you might conclude harms are probably way more common and more serious than any benefits.

But if you explained all that you’d be admitting that opinion has affected your recommendations. Neither the AAP nor the CDC seems prepared to do that.

But they should. Then, the ‘science’ would be evident, and accessible to all, and the judgment would be transparent. And people could use their own values to make decisions about the vaccine.

Imagine that.

Get full access to Research Translation at researchtranslation.substack.com/subscribe