This application seeks support to develop methods for the analysis of spatio-temporal area data. These methods aim to: identify trends in space and/or time; detect """"""""hot"""""""" or """"""""cold"""""""" spots of risk; assess delivery of interventions; detect populations who are subject to health disparities; and generate hypotheses concerning possible risk factors. Aggregated spatio-temporal disease data present unique statistical challenges, in particular one must account for: spatial and temporal dependence; the inherent instability of rare events; errors in numerators and denominators; and problems due to aggregation (which if not addressed may lead to """"""""ecological bias""""""""). This application addresses each of these issues, and has specific aims: 1. To develop and apply spatio-temporal models, in particular to carry out surveillance. 2. To develop a framework for analysis of the association between aggregated health outcomes and environmental exposures (in air, water, or soil) measured at point locations. 3. To develop efficient designs for combining ecological- and individual-level data to provide a stronger analytic basis for ecological studies. The methods developed will be genetically applicable to non-infectious diseases and will be illustrated using a range of data including publicly-available U.S. cancer incidence and mortality data, and Washington State small-area cancer incidence data. Developed methods will be made available to researchers via implementation within freely-available software.
Showing the most recent 10 out of 29 publications
