Modelling the impact of social mixing and behaviour on infectious disease transmission: application to SARS-CoV-2

TL;DR

This paper introduces a novel compartmental infectious disease model that incorporates socioeconomic and behavioural factors, such as social mixing and testing, to better understand and predict COVID-19 transmission dynamics in England.

Contribution

It develops a stratified Bayesian model accounting for deprivation and age, enabling real-time fitting and analysis of social behaviour impacts on disease spread.

Findings

Deprivation-related case trends reversed after summer 2021.

Differential social mixing explains changes in infection patterns.

Model can identify high-risk groups for targeted interventions.

Abstract

In regard to infectious diseases socioeconomic determinants are strongly associated with differential exposure and susceptibility however they are seldom accounted for by standard compartmental infectious disease models. These associations are explored here with a novel compartmental infectious disease model which, stratified by deprivation and age, accounts for population-level behaviour including social mixing patterns. As an exemplar using a fully Bayesian approach our model is fitted, in real-time if required, to the UKHSA COVID-19 community testing case data from England. Metrics including reproduction number and forecasts of daily case incidence are estimated from the posterior samples. From this UKHSA dataset it is observed that during the initial period of the pandemic the most deprived groups reported the most cases however this trend reversed after the summer of 2021. Forward simulation experiments based on the fitted model demonstrate that this reversal can be accounted for by differential changes in population level behaviours including social mixing and testing behaviour, but it is not explained by the depletion of susceptible individuals. In future epidemics, with a focus on socioeconomic factors the approach outlined here provides the possibility of identifying those groups most at risk with a view to helping policy-makers better target their support.