Early this week headlines proclaimed, “97,000 children tested positive for COVID-19 in the last two weeks in July.” Panic ensued in my inbox and Twitter feed.
“This means we cannot open schools!”
“Kids are actually the MOST likely to get it!”
“Help me! What does this mean?!”
The source of this fact is this (mostly) very helpful report the CDC put out, with all kinds of details. They are combining data from many states on pediatric cases, looking at case counts and hospitalization rates.
(Note: “children” is a broad definition here and varies by state. In most cases it is 0 to 17, although Alabama thinks of all people under 24 as children).
The top line number in the report is that, overall, based on positive tests we have so far, the COVID-19 rate in children is 447 cases per 100,000 children. This means that of every 100,000 children in the US, 447 of them have tested positive for COVID-19.
The report also includes a hospitalization rate, and a death rate. About 2% of cases are hospitalized, for a rate of 9 in 100,000 people. The death rate is 0.13 in 100,000.
The question is: what to make of all this? Here are some reactions, and some responses…
I thought kids didn’t get COVID-19, so this is really shocking.
Kids do get COVID-19. I think a number of policy-makers, including the President, have done a huge disservice by making claims that children do not get COVID-19. They can. But this isn’t news from this study.
I thought kids were less likely to get it
This is true, and not disproved by these data. If you dig into the CDC report you can see that in places which are doing a lot of testing (like New York), kids represent a small share of cases. This suggests they are less likely to get COVID-19 than older people.
But, honestly, these data are not the best way to learn these facts. We have lots of information out of Europe and elsewhere on kids and COVID-19 (check out this summary). The inconsistent testing, poor “child” definitions here, the fact that kids are often asymptomatic…all of this means that these numbers don’t really mean that much. The actual rate is probably higher for this whole age group (because we do not test enough asymptomatic people) and probably lower for younger kids. The data is just incomplete.
Can we learn anything?
For me, the most useful information out of this is the information on hospitalizations and deaths. Hospitalizations for COVID-19 are much better tracked than cases, and not nearly as subject to the testing issue. Looking at the hospitalization rate, we can say something about how likely children are to be seriously ill.
And, as I said above, the hospitalization rate is about 9 in 100,000 in these data. How can you contextualize that?
In the 2019-2020 flu season, the hospitalization rate for flu was 94 per 100,000 in children 0 to 4 and 24 in 100,000 in children 5 to 17. This averages to about 47 in 100,000 over the whole age range. This is about 5 times higher than COVID-19. This isn’t a hospitalization rate per case — it’s an overall rate, taking into account both likelihood of getting the disease and severity. Put simply: children were more than 5 times as likely to be hospitalized for flu over this period than for COVID-19.
The flu death rate for children per 100,000 is about twice as high as the COVID-19 death rate.
Kids may be more likely to get COVID-19 than the flu, but they are much less likely to become seriously ill from it. Not just less likely conditional on getting it, but less likely overall.
Okay, great, so that means that schools are fine? Or not fine? What?
In my view, these data are a nothing-burger in terms of schools. In a sense, they tell us things we already know: kids can get COVID-19 but it is typically not serious. But opening schools is about much more than this — it’s about transmission, adults transmitting to other adults, transit to school, etc. On top of the fact that there are benefits to school which are not (obviously) captured here.
Conclusion: please do not panic every time there is a scary headline. Breathe, read the context and think about what we learn.
