Leveraging Climate, Land Cover, and Health Monitoring to Develop a Malaria Early Warning System for the Amazon

William Pan , Duke University
Ben Zaitchik, Johns Hopkins University
Mark Janko, University of North Carolina at Chapel Hill
Carlos Mena, Universidad San Francisco de Quito
Francisco Pizzitutti, Duke University
Cristina Recalde, Johns Hopkins University
Andres Lescano, U.S. Naval Medical Research Unit No. 6

Malaria is a vector-borne disease causing an estimated 219 million infections and 435,000 deaths annually. Since 2011, no other region in the world has experienced a larger increase in malaria cases than the Amazon. Three factors have driven this malaria rise: strong ENSO events (2011-12, 2016); withdrawal of the Global Fund to Fight AIDS, Tuberculosis and Malaria; social unrest in Venezuela; and policies conducive to resource extraction, which increases both vector habitat and occupational migration associated with malaria transmission. These realities make malaria control challenging for health systems. Current surveillance and control programs rely solely on weekly case reports and respond to outbreaks with incomplete data (case reporting has a 1-4 week lag). Further, health response occurs in political districts, regardless of the strong environmental and demographic factors driving malaria risk, and independent of environmental policy. To address these challenges, we initiated the development of a Malaria Early Warning System (MEWS) in collaboration with the Peruvian government and support from NASA. Our MEWS forecasts outbreaks with 95% sensitivity and 75% specificity 12 weeks in advance in eco-regions (i.e., defined by demographic and environmental characteristics such as population, climate, hydrology, and land cover), and provides estimates of malaria incidence in small administrative districts with minimal error. The MEWS also includes agent-based models to simulate intervention response over short- and long-term time horizons. Our MEWS is the first early warning system capable of accurately forecasting health risks on a time scale that permits sufficient planning by the health system.

See extended abstract

 Presented in Session P999. Development, Environment and Space