Fredrik Lilijeros of the Karolinska Institute and Stockholm University, Petter Holme of the University of Michigan, and Johan Giesecke of the Karolinska Institute analyzed information about 295,108 in-patients at hospitals in Stockholm County during 2001 and 2002. They defined a "contact network" in which two patients were said to be in contact if they had been in the same ward at the same time.

"Traditional epidemiological modelling assumes that people meet with uniform randomness," Holme told PhysicsWeb. "Our study is an example of a rather new type of epidemiological modelling that takes the contact structure of the population into account. This is by far the largest contact network studied, and the conclusion that the epidemic thresholds for some diseases like MRSA and tuberculosis (TB) are lower than in traditional epidemiological modelling is new and unexpected."

The scientists say that their work could lead to better ways of detecting and controlling epidemics in health care systems in the future. They also plan to compare the predictions of the models with historical data.