Kids and School Transmission

The New York Times recently published this piece on school transmission, which suggested that there are serious risks to reopening schools (while also recognizing the fact that kids staying home for long periods may be detrimental).

To put this news in context, we will dig a bit into the papers they cover, and then look at what other research is out there on this question.

The Studies

There are two papers summarized in this piece. The first, which they cover briefly, is a study of viral loads in kids and adults. The authors show that among people with symptoms, viral loads in children and adults are similar. This paper is not peer reviewed, and has come under some criticism. It supports the view that children can transmit the virus, although it’s not a direct answer to whether they frequently do.

But in the NYT piece the main study was one published in Science. In this paper, the authors collect survey data from people in Wuhan on their number of contacts. In February 2020, they asked them about their number of contacts on the day before the interview and on a randomly chosen weekday in December 2019. They also collect similar data on Shanghai, although the “past” contact data in that case is based on an earlier survey from 2017–2018.

The authors then incorporate these data into an epidemiological model to predict the path of the epidemic under various scenarios. Focusing on kids, they incorporate two key assumptions into the model. The first assumption is that kids are less likely to carry the virus than adults; this is based on data on infection rates from other research. The second assumption is that conditional on being infected kids and adults infect others at the same rate. Based on these two assumptions, the model predicts that school opening would contribute a lot to virus spread.

The key intuition in their findings is that kids report more contacts than adults do. So even though kids are less likely to be infected, they will contribute a lot to the spread of the virus, assuming that they are equally infectious as adults. As others have pointed out, though, the assumption of equal infectivity is not supported by all data. One could also complain about retrospective survey data (i.e. can you recall how many people you saw on a random weekday two months ago?) and models are always subject to concerns about noise in the data.

The Context

What does the wider literature say here? If you want a very long-form version of this, check out our whole Kids and COVID Explainer. But we’ll hit a few highlights here.

  • The Lancet published a model-based study last month using evidence from (the original) SARS, the seasonal flu, and early COVID-19 data and suggested that schools were not likely to be an important source of spread. This is in contrast to the new paper in Science, and one could make similar complaints about the value of relying on models.
  • There are a few early pieces of data where researchers looked at cases of children or adults who have COVID-19 and analyzed whether they spread the virus in school settings. This includes:
  • There have also been attempts to look at spread within families.
    • Experts in Iceland report they have not seen any cases where children infect parents.
    • Researchers in the Netherlands have also been shedding light on this. They’ve been following families — preliminary data has 54 families with 239 people — and looking at infections. They have so far found no cases in which the child was the first one in a family to be infected.

Other Reading and What’s Next?

This piece in Nature has a good summary of the sides of the debate, and for another take on much of the above try here. And as for what’s next… schools in Europe are starting to open. As this happens, we’ll start to get a sense of what is happening with kids and their families. This will tell us a lot more about whether or not kids are an important vector for COVID-19.