Mobility measurement and network structure shape mortality inequalities in epidemic simulations

Digital trace mobility data are now widely used to infer infectious contacts and parameterize transmission dynamics within epidemiological simulations. However, mobility is not a single observable quantity and may be operationalized in mathematically distinct ways that exhibit different patterns of socioeconomic stratification. Using a metapopulation model applied to COVID-19 in Brazil, we examine how alternative operationalizations of mobility magnitude and network structure affect inferred epidemic mortality inequality dynamics. Specifically, we compare mobility measures based on the number of spatial units visited versus the probability of remaining within a spatial unit, as well as static versus dynamic mobility network formulations derived from multiple data sources, while holding remaining model inputs constant. Across model specifications, socioeconomic inequalities in mortality ultimately emerge, but their apparent timing differs systematically depending on how mobility is represented within the model. Simulations parameterized using mobility measures based on the number of spatial units visited and static mobility networks produced earlier emergence of mortality inequalities than simulations parameterized using stay-put probability measures and dynamic mobility networks. These differences arise because alternative mobility measures capture distinct socially stratified dimensions of mobility and dynamic mobility networks contracted and became more localized and assortative during the early pandemic period. These findings demonstrate that mobility operationalization constitutes a substantively consequential modeling choice capable of altering inferred epidemic inequality trajectories even under otherwise equivalent epidemiological conditions. More broadly, these results demonstrate that representational assumptions embedded within epidemiological models can shape inferred epidemic inequalities, bearing consequences for equity in infectious disease research and response.