Percolation across households in mechanistic models of non-pharmaceutical interventions in SARS-CoV-2 disease dynamics

Since the emergence of the novel coronavirus disease 2019 (COVID-19), mathematical modelling has become an important tool for planning strategies to combat the pandemic by supporting decision-making and public policies, as well as allowing an assessment of the effect of different intervention scenarios. A proliferation of compartmental models were developed by the mathematical modelling community in order to understand and make predictions about the spread of COVID-19. While compartmental models are suitable for simulating large populations, the underlying assumption of a well-mixed population might be problematic when considering non-pharmaceutical interventions (NPIs) which have a major impact on the connectivity between individuals in a population. Here we propose a modification to an extended age-structured SEIR (susceptible–exposed–infected–recovered) framework, with dynamic transmission modelled using contact matrices for various settings in Brazil. By assuming that the mitigation strategies for COVID-19 affect the connections among different households, network percolation theory predicts that the connectivity among all households decreases drastically above a certain threshold of removed connections. We incorporated this emergent effect at population level by modulating home contact matrices through a percolation correction function, with the few additional parameters fitted to hospitalisation and mortality data from the city of São Paulo. Our model with percolation effects was better supported by the data than the same model without such effects. By allowing a more reliable assessment of the impact of NPIs, our improved model provides a better description of the epidemiological dynamics and, consequently, better policy recommendations. •Bridging the gap between compartmental and household-level models for COVID-19.•Mean-field approach to modelling household contact patterns (percolation effect).•Useful for modelling interventions targeted at household-level.•Better agreement to data was observed when implementing the percolation effect.•General approach that can be applied to other dynamic infectious disease models.