Must Whatever Goes up Come Down? Mortality Gradients in the Emergence of COVID-19

According to fundamental cause theory, even as disease burdens shift, social inequalities in mortality persist. Yet for emerging infectious diseases, isolating the roles played by underlying social disadvantage from unequal access to protective interventions remains a challenge. This is amplified in the historically unique case of COVID-19: since mitigation measures were deployed contemporaneously with the disease's global dissemination, we lack empirical evidence to disentangle the effects of pre-existing social conditions from interventions. To address this gap, I developed a theory-guided, data-driven, geospatially explicit susceptible-exposed-infectious-recovered-deceased model of all 5,565 Brazilian municipalities. I parameterize pre-existing inequities through disparities in household transmission and infection fatality rates and model two interventions, non‑pharmaceutical interventions (NPIs) and vaccination, for which I simulate nine counterfactual distribution scenarios, from no interventions to real-world distribution and alternative more-equitable strategies. In the absence of interventions the model reproduced an "expected" gradient: mortality became increasingly concentrated in low socioeconomic status municipalities as the disease proliferated. NPI adoption following observed, socially stratified patterns accelerated, rather than created, this gradient, suggesting pre-existing social conditions as the primary drivers of Brazil's COVID-19 mortality gradient, while unequal interventions merely reinforced their effects. However, prioritizing the most vulnerable municipalities for vaccination inverted the gradient, an effect further compounded by equal adoption of NPIs, largely neutralizing the impact of these pre-existing conditions. These findings can help guide more equitable policy responses to future emerging infectious disease crises. This study is the first incorporating the pathways framework of inequalities in emerging infectious disease into mechanistic modeling.