Spatial heterogeneity and malaria dynamics in an urban environment
Urban environments contain pronounced heterogeneity in environmental conditions, population density and wealth, whose effects on infectious disease dynamics remain poorly understood. The spatial structure of these heterogeneities can potentially influence both susceptibility and exposure to infection as well as access to health care in vector-transmitted diseases such as malaria and dengue . As cities of the developing world continue to grow at unprecedented rates, it becomes increasingly important to better understand the role of spatial heterogeneity and the related spatial scales at which to aggregate the system to consider the influence of socio-economic and environmental factors, including climate variables. To address these questions for urban malaria, we take advantage of retrospective monthly case data for both Plasmodium falciparum and Plasmodium vivax infections resolved at the level of wards for the city of Ahmedabad, India, from 2002 to 2013. We analyse these spatio-temporal data sets with (1) a multiple regression model that incorporates population density, vegetation index (NDVI), temperature and socioeconomics indicators as covariates and (2) a dynamical approach that is spatially explicit and is formulated as an inhomogeneous Markov chain, adapted here for the purpose of urban malaria.
Despite considerable interannual variation in incidence, the city exhibits striking regularity in the locations with the highest malaria burden through time. This variation suggests the existence of two distinct regions, or groups of wards, whose transmission dynamics differ. With the statistical model, we identify a significant role of several concurrent factors including, population density, socio-economic conditions and environmental covariates, namely temperature and humidity. With the dynamical framework, we show the relevance of distinguishing two groups of wards, and a significant effect of temperature. These findings underscore the value of spatial structure in both environmental and population factors to predict malaria risk at city-wide scales.