A Challenge With Spatial Analysis and the Epidemiological Mapping of COVID-19 Cases
The built environment has long been conceived of as an instrument to manage disease. Rapid urbanization during the industrial revolution stressed and strained infrastructure, making cities hotbeds of disease and the sites of interventions. As a result, the field of public health, focusing on the science of preventing disease, developed concurrently with the rapid growth of the modern city.
John Snow’s (1813-1858) work with forensic mapping is considered fundamental to modern epidemiology and calls attention to the importance of understanding the urban geography of disease. Snow famously mapped the 1854 outbreak of cholera in Soho, London. Though the rising death count was apparent, the means of transmission of the disease and its source was not. Snow’s breakthrough was placing data in an appropriate context for analysis. Instead of a rising death toll, or a time-line, tabulating the growing deaths, he plotted each death on a map of central London. He created a spatial representation of the cholera outbreak, visualizing the diffusion of cholera, and ultimately, traced it to its origin, the Broad Street pump. He thus confirmed his hypothesis that the disease spread via the water supply. His analysis persuaded the authorities to remove the pump handle; the outbreak abated shortly after that.
I first saw Snow’s map and research in the context of using statistical and graphical reasoning in making life-and-death decisions, presented by Edward Tufte, Professor Emeritus at Yale University. Tufte’s stimulating presentation of Snow’s work and the implications for the use of statistical evidence comes to mind as I follow several dashboards mapping the spread of coronavirus across the globe.
One map I am following is Johns Hopkins University’s Center for Systems Science and Engineering. This dashboard lists confirmed COVID-19 cases, fatalities, and recoveries, mapping the occurrences globally (aggregating data from the World Health Organization; Centers for Disease Control and Prevention; European Centre for Disease Prevention and Control; among a growing number of state and national government health sources). As I follow different reports in the US, it is apparent that the perceived threat of the coronavirus is often expressed in relation to the number of established occurrences documented in the immediate area. In these cases, the map of confirmed cases is used to determine when the risk is significant in order to take action, proximal to an outbreak. This is flawed reasoning. It is problematic to use these maps of confirmed cases to determine when we should respond to the impending risk. Given the current lack of testing in the United States, by the time cases are identified, community transmission is most likely already a reality. Proactive precautions are better than reactions to effectively reduce the transmission of coronavirus and flatten the epidemic curve.
Today, robust geospatial information is allowing us to create sophisticated maps of the spread of coronavirus. However, thinking of Tufte’s thesis on the appropriate application of evidence for making life-and-death decisions, these maps may be better at identifying areas that have already taken prudent action when it mattered, not to signal when the level of the threat requires action. Ideally, we will be able to look to these maps to visualize how well different cities and regions are taking the necessary action (social distancing) at the critical time (before community transmission escalates).
Sources:
Martin Melosi. The Sanitary City: Urban Infrastructure in America from Colonial Times to the Present. Baltimore: John Hopkins University Press, 1999.
Charles Rosenberg. The Cholera Years: The United States in 1832, 1849, and 1866. 1962. Chicago: University of Chicago Press, 1987.
Edward Tufte. Visual Explanations, Images and Quantities, Evidence and Narrative. Connecticut: Graphic Press, 1997.
