Spatiotemporal relationships between extreme weather events and arbovirus transmission across Brazil

Brazil experiences large-scale annual outbreaks of dengue, chikungunya, and Zika virus infection, which are transmitted to humans by Aedes mosquitoes. Dengue is expanding into the north and south of Brazil and incidence of infection has been increasing over the last decade. Whilst previous analyses at the state and microregion level demonstrated that climate and environmental conditions affect dengue transmission, summary statistics computed over these large geographical areas can remove significant effects and mask local drivers of arbovirus transmission, which are needed for operational decisions on disease surveillance and control at the municipality level. We analysed the weekly case notification timeseries of chikungunya, Zika, and dengue virus reported at the municipality level for Brazil (n=5550 administration units) from 2013 to 2020 using spatiotemporal mixed-effects regression models. We used this highly granular data to test the association between arbovirus incidence and 139 variables capturing meteorological conditions, environment, El Nino Southern Oscillation (ENSO) patterns, human connectivity, and socioeconomics of the resident population. We assessed which factors best captured the historic transmission dynamics of chikungunya, Zika, and dengue including extremes of rainfall and temperatures at different time lags. Our findings highlight the joint health impact of poverty and extreme weather conditions on arbovirus infections in Brazil. We found that reduced socioeconomic indicators such as household income and access to adequate sanitation were associated with increased arbovirus incidence. Higher temperatures were positively associated with arbovirus incidence up to limiting negatively associated maxima. Summary statistics representing extreme conditions, such as the absolute maxima of environmental temperatures, ENSO anomalies, and long-term periods of extreme wetness or drought, were among the key predictors of arbovirus incidence. The findings presented in this study shed new light on the long-term drivers of dengue transmission at unprecedented spatiotemporal resolution, which in future work can be used to reconstruct the attribution of anthropogenic climate change and to evaluate how climate change scenarios are expected to affect arbovirus dynamics going forward. A version of this abstract in Portuguese can be found in the Supplementary Material.