Researchers use game to change how scientists study disease outbreaks

April 3, 2012

GAINESVILLE, Fla. — It may seem like a macabre game of tag, but it’s actually an innovative tool for teaching the fundamentals of epidemiology — the science of how infectious diseases move through a population.

University of Florida biologist Juliet Pulliam is among an international team of scientists who teach a workshop annually in South Africa that helps epidemiologists improve mathematical models they use to study outbreaks of diseases like cholera, AIDS and malaria. Pulliam and colleague Steve Bellan from the University of California Berkeley created the game in 2010 as a teaching aid for the workshop. The exercise has proven so effective in demonstrating concepts in epidemiology that a discussion of the game is presented in the April 3 edition of the journal PLoS Biology.

In the game they call “Muizenberg Mathematical Fever,” players simulate a real-life epidemic by passing around pieces of paper that say, “You have been infected,” followed by instructions for propagating the disease. The drill has become a cornerstone of the annual training they offer at the African Institute for Mathematical Sciences in Muizenberg, South Africa.

“Infectious disease modeling is an established field of study in bio-mathematics,” said Pulliam, a biologist at UF’s Emerging Pathogens Institute and co-author on the paper. But there has been a tendency in the past for mathematicians doing that sort of work to operate separately from practitioners on the ground who track diseases as they are spreading, she said. The game was meant to convince all players on the epidemiology field that teamwork is the better approach.

“Not knowing how data about an outbreak was collected can lead to misinterpretations,” Pulliam said. For example, if procedures change for how infected individuals are counted, it could create a spike in the data that falsely portrays how the disease is actually being spread. The misinformation, once introduced into a model, could throw off projections and interfere with efforts on the ground to prevent further outbreaks.

Bellan, lead author on the paper, cites cases where collaborations between bio-mathematicians and classical epidemiologist have resulted in valuable lessons for tracking the spread of diseases.

HIV interventions and efforts to eliminate trachoma, a bacterial infection that causes blindness, have successfully used the tag-team approach, he said. In both cases, studies have shown that when practitioners employ the power of mathematical modeling to improve their intervention strategies, they are more likely to interrupt the progress of an epidemic.

“This is about the importance of collaboration,” said Bellan, an ecologist who specializes in epidemiology of wildlife diseases. “No one can be an expert in everything.”

“In fact, the two sides typically meet up somewhere along the line during the process of an epidemiological study,” he said. “We just want to see more scientists working together from the start.”

To that end, Bellan, Pulliam and six scientists from South Africa, Canada and the United States, offer two-week clinics every year at the African Institute for Mathematical Sciences. The clinics are meant to immerse epidemiological number crunchers more fully into the human aspects of how disease spreads. John Hargrove, former director of the South African Center of Epidemiological Modeling and Analysis, started the clinics in 2006. But since 2010 classes have had a very different experience than previous groups, because the game changed everything.

“We were sitting in the office the night before the clinic began, talking about how someone had shown up sick one year and gotten everyone else sick,” Pulliam said. “And then Steve said something about how cool it would be if we had captured data from that outbreak for use in the workshop.”

The discussion sparked the idea to create a similar scenario in real time by creating a fictitious disease. “We pieced it together in about an hour,” Bellan said.

An “infectious” piece of paper serves as the agent for spreading “Muizenberg Mathematical Fever.” The paper notifies the receiver that they have been exposed and then instructs the infectee to email Bellan of their fate, use a random number generator to determine how many others should be infected, and then pass the appropriate number of papers to other participants at the clinic.

The rules serve to propagate the disease but also to build a data set of who infected whom and when. “There were a lot of sneaky smiles going around that day,” Bellan said.

Pulliam said, “The drill produced an outbreak with data that looks like a real epidemic.” Pulliam said.

Clinic attendees are usually more mathematician than epidemiologist, she said. They typically spend the first week just talking about where data sets come from, who collects them, and what the numbers refer to. “Just using the game as a way to demonstrate those issues instead of talking about them is instructive on its own,” she said.

But the real benefit comes during the second week, when people working in groups experiment with various epidemiological models using actual data sets — typically from HIV studies or other ongoing projects.

“Many opted to work with data sets from the game,” Pulliam said. “Because they were really tangible.” They found that familiarity with the process for collecting data greatly improved their ability to customize mathematical models so that they accurately represented how a disease was moving through a population.

“And that’s exactly what we wanted them to get out of the workshop,” she said.