Spatial heterogeneity in transmission and the impact of interventions: a mathematical modelling approach

This COVID-19 Rapid Response award is jointly funded (50:50) between the Medical Research Council and the National Institute for Health Research. The figure displayed is the total award amount of the two funders combined, with each partner contributing equally towards the project. Predicting the size and duration of potential COVID-19 outbreaks is an essential component of public-health planning and preparedness. Mathematical models of disease transmission are potentially powerful tools for predicting the course of an upcoming epidemic and evaluating control and mitigation strategies. However, standard models of disease transmission without population structure overestimate the speed of invasion of a novel pathogen. We have developed a spatial metapopulation transmission model for the UK that is grounded in demographic data which incorporates regular (commuter-like) movements of individuals. In previous work, we demonstrated that regular, repeated movements lead to slower epidemic spread. Adapting this model for COVID-19, we estimated that an uncontrolled epidemic in England and Wales would peak ~4 months following sustained person-to-person transmission, but that seasonality in transmission could substantially alter the timing and magnitude of the peak burden. Here, we propose to use this model to evaluate control and mitigation strategies for COVID-19. Guided by the World Health Organization-identified research priorities and PHE needs, we will estimate the impact of travel restrictions, border screening and quarantine policies. We will also assess the effects of social distancing measures and other non-pharmaceutical interventions on peak burden and epidemic timing and rank measures by effectiveness. The model will also be adapted to assess and rank pharmaceutical deployment strategies. Our vision is to make the model adaptable and available to other countries and settings, both with and without census and commuting data. Key challenges include modelling commuting patterns, incorporating realistic age structure, adding an observation model to capture morbidity and mortality and including behaviour change which could substantially alter dynamics.