A conceptual framework for visualizing seasonality in spatiotemporal health data using dynamic maps

Recent advances in Geographic Information Systems (GIS) and visualization software allow for the representation of public health data over space and time through dynamic maps. Animation has promising potential in showing the spread of infectious disease in relation to climatic cofactors at various spatial and temporal scales. However, dynamic maps represent a delicate balance between being useful versus incomprehensible from information overload. In this study, we develop a conceptual framework for building informative dynamic maps using infectious disease and climate data. Designing dynamic maps requires several considerations beyond the traditional cartographic representations for static maps including: choice of a temporal scale, cartographic representation of environmental covariates with disease outcomes, appropriate symbolization of disease outcome data to demonstrate anomalies across all time periods, selection of the duration and rate of change for each map frame, and tools to direct the audience's attention to space-time variations and anomalies. Monthly Salmonella hospitalization data for the U.S. elderly were abstracted from the Centers for Medicare and Medicaid Services (CMS) MedPAR data file for a 5-year period (1998-2002). Monthly rates were calculated for each county and mapped using graduated dot symbols that represent the entire distribution of the data over all time periods. Salmonella hospitalization rates for each month were overlaid onto average monthly maximum temperature data from the PRISM climate data set. The resulting dynamic map reveals seasonal patterns in Salmonella hospitalization during warm summer months and suggests that dynamic mapping is informative for displaying the spread of infectious disease through time and space.