Comparing Various Methods for Sentinel Surveillance Site Placement

• Activity Number: 336
• Type: Topic Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Geoffrey Fairchild*+ and Alberto Segre and Philip Polgreen and Gerard Rushton
• Companies: University of Iowa
• Keywords: influenza ; surveillance ; geography ; algorithm ; sentinel ; disease

Abstract

Background: Influenza-like illness data is collected via an influenza sentinel surveillance network at the state level. In order to gather the best disease spread statistics, placement of these sites is important. Methods: We implemented two surveillance site placement algorithms; one maximizes population coverage while the other minimizes average distance. Using influenza data culled from Medicaid billing data, we evaluated the effectiveness of our algorithms compared to the sites hand-picked by the Iowa Department of Public Health (IDPH). We analyze different subsets of ICD-9 codes to determine which are necessary for peak disease detection. We also analyze the temporal aspects of disease spread to determine if a dynamically changing surveillance system would prove beneficial. Results: Simulating the spread of influenza across the state of Iowa, we show that our sites chosen algorithmically outperform the IDPH's hand-picked sites in terms of number of cases detected. Conclusions: Using our models, we can algorithmically place disease surveillance sites across a region in order to gather the best disease spread statistics.