Wednesday, December 1, 2010

Case-Control Geographic Clustering for Mobile Populations: Methods for Infectious and Chronic Diseases

Several years ago, my Dad and I wrote a paper on the best way to study infectious and chronic diseases. Because people live in different locations during their life and spend a large portion of their life at work, spatial and residential histories must be gathered and implemented in time based software. The case for this type of methodology is made by illustrating its effectiveness with a bladder cancer study conducted in South East Michigan. Not only did I get to help write this paper, I got to collect water samples from the case and controls homes and work (2 samples, one frozen and one not), urine samples and toenail samples. Thankfully, I did not have to work with the samples in the lab, but reaped the benefits. Our paper can now we be found online (abstract below)!

Humans are mobile and constructs of Geographic Information Science have been used
to model daily and weekly activity patterns, as well as residential and work spaces. But
geographic epidemiology often ignores human mobility and employs methods that
assume humans are sessile rather than mobile. This paper first quantifies how relaxing
the assumption of sessile individuals might impact case-control cluster tests, and finds
the results are highly sensitive to when the system is observed. Recently developed
tests for case-control clustering that account for human mobility are then presented,
along with extensions to the analysis of infectious disease data. We conclude by
revisiting an analysis of bladder cancer in south eastern Michigan, and demonstrate the
ability of the new techniques to detect global and local clustering in case-control data
for residential histories. Statistical techniques that account for human mobility are
needed for chronic and infectious diseases where causative exposures occur at
locations different from ones location at time of diagnosis.

No comments:

Post a Comment