Answering Tough Flu Questions

The 2009 H1N1 flu epidemic raised many questions for public health officials. Using TeraGrid supercomputers, researchers modeled the flu outbreak and helped policymakers locally, nationally and internationally evaluate strategies for responding to H1N1 in real-time during the epidemic.

Epidemiological modeling--using computational tools to mimic how infectious diseases spread through populations--can help answer both medical and social questions. Modeling can examine issues such as whether schools should be closed and if so, for how long? If only a limited amount of vaccine is available, which groups--children, elderly or caregivers--should be vaccinated first? With limited supplies of antiviral medication, which communities should get these relatively new medications and in what quantity?  

One researcher, Lauren Ancel Meyers of the University of Texas at Austin, used the Texas Advanced Computing Center's Lonestar system to model H1N1 transmission within and among U.S. cities. Her model optimized choices for how to distribute the 50 million doses of antiviral medication. This work showed that relatively simple strategies--such as regular releases of the stockpile to cities in proportion to their populations--worked as well as more complex approaches. Her findings are especially pertinent for the future, since they indicate that antivirals can save lives and reduce transmission prior to the availability of a vaccine.

Shawn Brown and Daniel Burke of the University of Pittsburgh's Graduate School of Public Health's MIDAS Center of Excellence collaborated with Meyers. They used the Pople Computing System at the Pittsburgh Supercomputing Center to model H1N1 on a regional basis and in metropolitan areas. Their models are a virtual laboratory to ask questions you can't ask with real populations. They found that to close schools less than two weeks may slightly increase infection rates and that schools may need to be closed eight weeks or longer to have a significant impact.

Their modeling also supported recommendations that priority be given in vaccinations to people at risk for severe complications. They found that prioritizing at-risk individuals, may result in slightly more cases of flu, but reduces serious disease and death, and overall economic cost. Bruce Lee of the University of Pittsburgh Graduate School of Public Health also collaborated on the project.

A number of U.S. epidemiological modeling research groups share their knowledge through the National Institutes of Health's MIDAS (Models of Infectious Disease Agent Study) program, which helps health officials prepare for outbreaks.