Diseases may be aggregated in certain space (geographical clustering) or in certain time periods (time clustering). Several statistical methods have been developed to determine disease clustering in space, time, and space and time. In this study, Grimson, Besag-Newell, Kulldorff’s spatial scan, Lawson-Waller score methods used in space clustering of disease were promoted. Findings that were obtained from the applications of some methods were comparatively examined.
Application data were obtained from 248 patients under treatment for nasopharynx cancer in Hacettepe University Oncology Hospital, Radiation Oncology Department between October 1993 and June 2002.
Global spatial disease clustering wasn’t statistically significant according to Grimson’s method. When local spatial clustering methods were applied, local disease clustering was found significant with all methods, in different areas. In Kulldorff’s spatial scan test, different results were obtained by selecting different scanning window sizes. When scanning window size is increased, the method became more sensitive to large clusters. In Besag and Newell method, local clustering were not significant for k=2, 4, and 6, and local clustering were significant for k=8, 10, 12, 14 and 16.
Although there was no evidence of focus of risk in our data, in order to examine the disease clustering around a focus by Lawson and Waller score test, the coordinates of Çankaya-Ankara was set as the coordinates of a focus. Clustering was found significant according to Lawson and Waller score test around this focus.
Key Words: Disease clustering, Space clustering, Spatial pattern, Nasopharyngeal carcinoma.
|