Scan statistics modeling for Crohn's disease clusters


What: Two models for detecting clusters of CD, based on scan statistics.

Why: Crohn's Disease(CD) is an inammatory disease of the intestines, which has no known pharmaceutical or surgical cure. In addition, geographical variations of Crohn's Disease incidence have been reported worldwide reacting putative variations in the distribution of environmental factors. The scan statistics is a statistical method designed to detect a local agglomeration of events and to test the randomness of such an agglomeration. First model supposes that the population is aligned in a sequence of iid Bernoulli f0; 1g random variables (not taking into account the geographical component), and gives two approximation formulas for the distribution of scan statistics (Naus 1982 and Haiman 2007), the second with an error bound. The second model is based on the spatial and space time scan statistics (developed by Kulldor_ 1997), under the same assumption that the number of CD cases in each canton is binomial distributed. We have used the SatScan software. A numerical comparison is given between the two models.