The role of super-diffusion in the coronavirus pandemic

A mathematical analysis shows that super-diffusion events have a much greater weight in coronavirus transmission than previously predicted. The study on Pnas

(photo: Equinox Graphics / Mit) From sporting events to ceremonies. Super-spread events, i.e. when one coronavirus positive person infects many others, have a huge weight in the overall spread of Covid-19. This was reported by researchers from the Massachusetts Institute of Technology (MIT), who through specific mathematical and statistical analyzes have shown how super-spreading, or to put it in English superspreading, has a much greater impact than expected. Their study has just been published on the pages of Pnas.

To understand this, the researchers considered about 60 super-diffusion events, 45 recorded by the current pandemic and another 15 documented in the literature and dating back to the epidemic of 2003, showing that the occasions when a person infects at least 6 others are much more common than expected. In fact, the data showed that in most of these events, between 10 and 55 people were infected, while in two, both in 2003, more than 100 people were infected.

As the researchers explain, for the coronavirus the so-called basic reproduction number, R0, which indicates the potential transmissibility of an infectious disease, is about 3: that is to say, therefore, that each person infected with the coronavirus will spread it to three other people on average. However, this parameter varies greatly from person to person, as some subjects do not spread the disease to anyone else, while others, the super-diffusers, can infect dozens of people (in the new study, the latter were defined as individuals who transmit the virus to more than six people). Given the commonly used statistical distributions in which a positive infects three other individuals, events in which the disease is transmitted to dozens of people are considered extreme, very unlikely. For example, a “normal” distribution might look like a kind of bell with a peak close to three and with a tail that tapers quickly in both directions. In this scenario, therefore, the probability of an extreme event decreases exponentially as the number of infections moves away from three.

But this is not the case with coronavirus super-spread events: using tools mathematicians of the theory of extreme values ​​(a branch of statistics applied in the world of finance), the researchers were able to quantify the risk of so-called "fat-tail" events (situations that form a large rather than thin tail) and to demonstrate that even if super diffusion events are extreme, however they are likely to occur. “We have shown that the likelihood of extreme events decays more slowly than would be expected,” explain the researchers. "Large super-diffusion events - between 10 and 100 infected people - are much more common than we anticipated."

In addition, researchers have developed a new mathematical model of transmission of Covid-19, showing that limiting encounters to 10 people, or even fewer, could significantly reduce the number of super-broadcast events and lower the total number of coronavirus infections. "Super-broadcast events are probably more important than most of us initially assumed," comments James Collins, co-author of the study. "If we can control them, we have a much better chance of managing this pandemic." One way to do this, the researchers conclude, would be to prevent anyone from interacting with more than 10 people at the same time.