Modeling malaria risk in the Peruvian Amazon

Malaria continues to be one of the world's most devastating threats to public health. In Peru, 75% of malaria cases occur in the Department of Loreto, in the Northern Amazonian Region, and most cases (80%) are concentrated in just 10 of the department's 51 districts. Key factors related to continued malaria endemicity in this region are the expansion of vector habitats from land use change (deforestation for logging and road development) under humid tropical climate conditions and social and ecological processes that increase human exposure to the Amazon's most efficient malaria vector, Anopheles darlingi. In order to refine and focus prevention strategies across the region, spatially explicit risk estimates are required. In this study, we investigate malaria risk across time and space in this region by modeling: (1) the influence of land cover, land cover change, and hydrometeorological conditions on the distribution of An. darlingi and other malaria vectors, and (2) the relationship among climate, land use, and confirmed malaria case counts at regional health posts from 2006 to 2011. Land cover analysis is performed using Landsat imagery in combination with supplementary data, and meteorological and hydrological conditions are simulated using a customized Peruvian Amazon Land Data Assimilation System (LDAS) that makes use of satellite observations from TRMM, AMSR-E, and MODIS and the Noah Land Surface Model. Mosquito distributions are measured in situ, while malaria risk is evaluated by estimating populations associated with delineated geographic catchment areas for each of the health posts. Here we present results of the mosquito and case prediction models. It is found that information on percent forest cover, near surface soil moisture, maximum daily temperature, precipitation, and other hydrometeorological variables contribute to skillful prediction of An. darlingi counts. For malaria case prediction, these variables and longer term climatic averages contributed to model skill. These results inform the ongoing implementation of a malaria risk assessment and early warning system for the region that will enable our local partners to implement more efficient prevention and control efforts.