Title: Incorporating population-at-risk uncertainty into disease mapping models
Authors: Lance Waller - Emory University (United States) [presenting]
Abstract: Small area health studies typically rely on multiple sources of data to define local disease risk with health registries providing information on observed outcomes of interest and census (or other administrative) data defining the numbers of individuals of risk and potential sociodemographic risk factors. In the United States, some demographic aspects (e.g., age, race, and sex) are available for census small areas (tracts, block groups, and blocks) from the U.S. Census Short Form, while some (e.g., economic variables, housing) were historically available for census small areas from the U.S. Census Long Form which was replaced by the American Community Survey (ACS) in 2010. The ACS provides a rolling sample of the U.S. population and provides small area estimates for specific time periods and with measures of error. We provide a brief review of the role of census demographics in small area health studies, define the available data, and illustrate the impact on small area health studies using data from the 2000 and 2010 U.S. Census and small area health statistics from the state of Georgia, with particular attention on incorporating the reported ACS error into typical models of small area health effects.