Do Coronavirus Superspreaders Exist?

A number of articles in the past week or so have focused on the possibility of COVID-19 “superspreaders.” This is the idea that a very large share of COVID-19 infections are caused by a small number of people. For example, one article from the Telegraph suggests that perhaps 80 percent of infections are caused by 10 percent of people. Is this true? And if so, you may be wondering: How can I avoid that 10 percent of people?

To think about this, we need a little bit of epidemiology. If you’ve followed anything about COVID-19 carefully you’ve probably heard about “R0,” the effective reproductive rate of the virus, which measures the average number of people that are infected by one infected person. If R0 is larger than 1 (i.e., if, on average, an infected person spreads the virus to multiple people), the number of infected people continues to grow.

There is also a second, less discussed epidemiological parameter: k, the “dispersion factor.” This k is a measure of the concentration of infection, of its clustering. It’s a way to get at the question: Do we see a lot of dense clusters of infection, or are the patterns more dispersed? This parameter doesn’t have a direct interpretation like R0, but what we can say is that a lower value of k implies a greater clustering of infections. Greater clustering implies that a smaller number of people are responsible for more infections.

If you want to get into details, Science had a nice general write-up on this topic, and here is a (denser) preprint about estimating this k value in COVID-19. (There has been a lot of discussionof the value of preprints—early release, non-peer-reviewed studies—in COVID-19 research. On the one hand, they are fast. On the other hand, lack of peer review may limit our confidence in results. It’s a balance.) The preliminary data suggest that COVID-19 could have a very low k; perhaps something like 0.1. If that’s the case, then a small number of infected individuals could be responsible for a large number of infection clusters.

Keep reading