Clinical Utility of Preemptive Surge Measures Enacted as a Result of Conventional and Internet-Based Influenza Surveillance 






Will coupling either traditional and/or novel internet-based influenza surveillance signals with astrategically focused response plan result in significant improvements in ability to match care demands with resources?

Our rationale for the work: In spite of billions of dollar invested there is a conspicuous absence of studies demonstrating improvedoutcomes in medical systems as a result of syndromic surveillance. Internet based surveillance offers multiple advantages overtraditional surveillance including real-time, easy to access data, improved geographic details due to a wider population base, and mostimportantly, earlier warning (Ginsberg et al, Nature 2009); applications are particularly relevant to acute care settings such asemergency departments (EDs) where unplanned surge negatively impacts patient and public health outcomes.


Completed Project Overview


Year 1: We achieved the following: established working relationship with Google Flu Trends (GFT); developed infrastructure for datacollection from multiple complex systems; conducted first-time ever study demonstrating local validation of GFT including association withED crowding indices. GFT strongly correlated (with near real-time) with both ILI and confirmed influenza and preceded select crowdingmeasures (Dugas et al CID 2012). This work sparked multiple high level press releases (including NPR and TIME) and invitations toconferences (including Intl Conference on Digital Disease Detection at Harvard University).

Year 2: We conducted an expert panel with key regional and institutional stakeholders to develop a feasible strategic plan to link withinfluenza surveillance data thereby operationally integrating advanced warnings to manage ED surge and improve public health outcomes(Dugas et al PLOS Currents Disasters 2013). Additionally, we built on the findings from year one to demonstrate GFT's predictive (vs.correlative) capacity for both influenza and ED crowding through the use of Generalized autoregressive moving average (GARMA)modeling (Dugas et al. PLoS One 2013)

Year 3: We consolidated findings from years 1 and 2 to create a prototype web-based tool for hospitals to help anticipate, plan, and respond to the surge associated with both seasonal and pandemic influenza. This web-based tool automatically interfaces with GFT dataand permits individual EDs bot regionally and across the United States to input their personalized data and forecast future influenza surge attheir institution. We built the key federal relationships necessary to expand the validation of GFT and the corresponding GARMA model to multiple cities, via the Healthcare Cost and Utilization Project (HCUP) database through partnership with leadership at the Agency forHealthcare Research and Quality (AHRQ). Additionally, we enrolled 300 patients in a study to evaluate the accuracy of current surveillancemethods, such as confirmed influenza cases from routine clinical testing, number of patients presenting with ILI, and GFT, relative to theactual rate of influenza positivity among ED patients determined by systematic influenza testing.

Register and use the web-based tool FluCast

Principal Investigator(s) and Researchers(s)

Richard E. Rothman, M.D., Ph.D.

Principal Investigator, PACER