COS 132-8 - Aggregating spatial disease data: When high-resolution matters most

Friday, August 12, 2011: 10:30 AM
10B, Austin Convention Center
Nita Bharti, Ecology and Evolutionary Biology; Woodrow Wilson School of Public and International Affairs, Princeton University, Princeton, NJ, Matthew J. Ferrari, Biology, Penn State University, University Park, PA and Bryan T. Grenfell, Ecology and Evolutionary Biology; Woodrow Wilson School of Public & International Affairs, Princeton University, Princeton, NJ

Understanding the spatial progression of recurrent epidemics can improve disease management. For density-dependent, directly transmitted diseases, spatial waves of infection often travel down population density gradients. These waves can be detected from sufficiently detailed spatial data and are seen in human infections, most notably for historical and recent epidemics of measles and influenza from industrialized nations. These patterns of epidemic progression reflect general patterns of human movement, which are relatively well characterized in the developed world. In contrast, many areas with high current disease burden are found in low-income nations where human movement is poorly understood. These disease data are often aggregated over large areas and various levels of host density, losing spatial resolution and obscuring host density gradients. As a result, the spatial progression of coarsely reported epidemics can be difficult to discern. This may be due to an absence of structured spatial epidemic progression or could result from the inability to detect spatial patterns of outbreaks from low-resolution disease data. To determine the required data resolution at which spatial patterns of incidence can be detected, we increasingly aggregated a high-resolution measles incidence time series from pre-vaccination England and Wales, shown to display hierarchical spatial waves of progression, until it matched that of reported measles cases from recent epidemics in West Africa.


We measured the relative impact of data aggregation on our ability to detect spatial waves of infection by aggregating incidence across 1) varying levels of population density and 2) over varying areas of relatively constant density. Overall we found that aggregating data across highly heterogeneous population density strongly concealed the spatial progression of epidemics. In other words, high-resolution disease reports over evenly populated areas do not greatly improve the detection of epidemic progression while high-resolution incidence data from cities and their surrounding areas greatly benefit the detection of spatial patterns of epidemics.

Copyright © . All rights reserved.
Banner photo by Flickr user greg westfall.