Power comparisons for disease clustering tests
Computational Statistics & Data Analysis
Gaussian Markov Random Fields: Theory And Applications (Monographs on Statistics and Applied Probability)
Spatial scan statistics in loglinear models
Computational Statistics & Data Analysis
Clustering for the localization of degraded urban areas
ICCSA'12 Proceedings of the 12th international conference on Computational Science and Its Applications - Volume Part II
| Hi-index | 0.00 |
The classical likelihood ratio spatial scan statistics has been widely used in spatial epidemiology for disease cluster detection. The question is whether the geographic incidence pattern is due to random fluctuations or the map reflects true underlying geographical variation due to etiologic risk factors. The hypothesis underlying the classic scan statistics assume that disease counts in different locations have independent Poisson distribution; unfortunately, outcomes in spatial units are often not independent of each other. Risk estimates of areas that are close to each other will tend to be positively correlated as they share a number of spatially varying characteristics. Ignoring the overdispersion caused by spatial autocorrelation leads to incorrect results. To overcome this difficulty, we propose a model-based approach adjusting for area-specific fixed-effects measuring potential effect modifiers, and for large-scale geographical variation of etiologic factors that vary continuously in space and are not expressly present within the model. We apply our methodology to the spatial distribution of lung cancer male mortality occurred in the province of Lecce, Italy, during the period 1992-2001.