The Bayesian Aerosol Release Detector (BARD) is a system designed to detect and characterize disease outbreaks caused by aerosol releases of B. anthracis. The detection algorithm of BARD requires, among other things, an accurate estimation of the number of spores that would be inhaled under a specific release scenario. This is a challenging problem, in part due to the lack of fine-grained data on the mobility patterns of the exposed population. Indeed, the only type of spatial information routinely contained in biosurveillance databases is the residential administrative unit - such as the home zip code - of each case. The current version of BARD detector deals with this challenge by making the simplifying assumption that exposure to anthrax would occur at one's residential unit. This paper presents an experimental study to assess how BARD's performance would be impacted by incorporation of a commuting model in outbreak simulation. Our results show that incorporation of commuting in simulation leads to statistically and practically significant changes in BARD's detection and characterization performance.