Choice Works

PSC Models Show Value of Vaccine Choice in Fighting Flu

Less than half of children and adults under age 65 get vaccinated for influenza. Researchers at PSC, the University of Pittsburgh and Soongsil University in the Republic of Korea used PSC supercomputers to simulate the effects of offering different types of vaccination—the familiar injected vaccine or two types of “needle sparing” vaccines. Their results suggest that such a choice would reduce flu cases and make vaccination more cost effective.

Why It’s Important

Each year, influenza hospitalizes about 226,000 Americans and kills an average of 24,000. One study estimated an economic annual cost of flu sickness to be over $85 billion. But less than half of children and adults under age 65 actually get vaccinated.

Jay DePasse of PSC’s Public Health Applications Group and collaborators at PSC, the University of Pittsburgh and Soongsil University in the Republic of Korea used PSC supercomputers to find out whether offering different types of vaccination—the familiar injected vaccine or two types of “needle sparing” vaccines—would reduce flu cases and make vaccination more cost effective. Previous work by the group using simpler models had shown that offering adults and children a choice of vaccines would increase the number vaccinated, but not how effective or cost-effective that increase would be.

“The latest paper caps a series of studies that have looked at the question of vaccine choice from the point of view of simple statistical models to more sophisticated ‘agent-based’ models. The increased computational power of the Bridges system allowed us to build on our earlier work with a massive, agent-based simulation.” —Jay DePasse, Pittsburgh Supercomputing Center

How PSC and XSEDE Helped

Using Olympus, PSC’s dedicated public health supercomputer, the scientists first determined that vaccine choice would, on the average, reduce cases and bring down the cost per dose of vaccination in the Washington, D.C., population. They built on earlier, simpler models using a tool called agent-based modeling (ABM). This method simulates individual people in an area as they go to work or school or socialize, watching how the virus spreads and how vaccination affects that spread. In a report in the American Journal of Epidemiology, they showed that offering vaccine choice reduced flu cases in both adults (who got the choice of a traditional injection versus a very-small-needle intradermal injection) and children (whose “virtual parents” chose between traditional and inhaled intranasal vaccines), and that the decreased costs due to illness offset the extra expense of offering more vaccines, reducing the overall societal cost of influenza. While offering choice to adults and to children separately helped, choice for both groups provided the best protection and lowest costs.

The group then moved their computations to the larger, XSEDE-allocated Bridges system at PSC, enabling them to expand their model to multiple regions of the country with very different populations, including Allegheny County, Pa.; Wayne County, Mi.; Santa Clara County, Calif.; and Salt Lake County, Utah. The new work reproduced the results seen in the D.C. study, and Bridges’ power allowed the scientists to test a wider range of assumptions about increased coverage and virus spread, showing that even moderate increases in coverage due to offering more choices can reduce costs and decrease influenza cases by 5,600 to 35,000 people across all five counties. The new work appeared in the journal Vaccine in July, 2017. The group is now continuing their work on an upgraded Olympus, which has incorporated many of the hardware innovations of Bridges.

“One of the reasons we used an agent-based model is it tests the indirect effects. Say I get vaccinated; that has an impact on whether or not my kids are going to get the flu. But it’s computationally expensive … We were able to say, ‘Well we have the computational power, let’s go for it.’ That adds to the realism of the simulation.” — Jay DePasse, Pittsburgh Supercomputing Center