Estimating the population-level effects of non-pharmaceutical interventions when transmission rates of COVID-19 vary by orders of magnitude from one contact to another

TL;DR

This paper models COVID-19 transmission considering power-law variability in contact transmission rates, estimating that widespread mask use can significantly reduce the effective reproduction number.

Contribution

It introduces a statistical physics-based model to estimate population-level effects of interventions accounting for transmission rate variability.

Findings

Mask-wearing reduces the reproduction number by a factor of about nine.

Transmission probability follows a power-law distribution across contacts.

Modeling variability improves understanding of intervention impacts.

Abstract

Statistical physicists have long studied systems where the variable of interest spans many orders of magnitude, the classic example is the relaxation times of glassy materials, which are often found to follow power laws. A power-law dependence has been found for the probability of transmission of COVID-19, as a function of length of time a susceptible person is in contact with an infected person. This is in data from the United Kingdom's COVID-19 app. The amount of virus in infected people spans many orders of magnitude. Inspired by this I assume that the power-law behaviour found in COVID-19 transmission, is due to the effective transmission rate varying over orders of magnitude from one contact to another. I then use a model from statistical physics to estimate that if a population all wear FFP2/N95 masks, this reduces the effective reproduction number for COVID-19 transmission by a factor of approximately nine.