The quantitative evaluation of pathogen transmission in the medical intensive care unit (MICU) is difficult given the small number of patients and the complexity and severity of illness. We sought to evaluate the suitability of a probabilistic computer model of our MICU, with which we could rapidly simulate infection control measures and other clinical interventions that would be impossible to perform in the real clinical setting.

A functional model based on observed behavior and work patterns of nurses and length of stay of patients in our MICU was implemented using two simulation software applications. The resulting epidemiologic data was compared for the purposes of validation. Variations in the random number generators across applications caused significant differences in simulation outcomes. The application that performed more realistically produced valid results that efficiently and accurately modeled the real MICU.

Performing simulations in silica is advantageous, especially in small-population clinical care settings such as the MICU. Identifying simulation tools and strategies that best fit the population being studied is an essential step and one that can have a large impact on the validity ofresults.