Examining the impact of the number of regions used in cluster detection methods: An application to childhood asthma visits to a hospital in Manitoba, Canada

Submitted: 10 April 2018
Accepted: 30 September 2018
Published: 9 November 2018
Abstract Views: 1141
PDF: 521
APPENDIX: 244
HTML: 14
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Authors

The level of spatial aggregation is a major concern in cluster investigations. Combining regions to protect privacy may result in a loss of power and thus, can limit the information researchers can obtain. The impact of spatial aggregation on the ability to detect clusters is examined in this study, which shows the importance of choosing the correct level of spatial aggregation in cluster investigations. We applied the circular spatial scan statistic (CSS), flexible spatial scan statistic (FSS) and Bayesian disease mapping (BYM) approaches to a dataset containing childhood asthma visits to a hospital in Manitoba, Canada, using three different levels of spatial aggregation. Specifically, we used 56, 67 and 220 regions in the analysis, respectively. It is expected that the three scenarios will yield different results and will highlight the importance of using the right level of spatial aggregation. The three methods (CSS, FSS, BYM) examined in this study performed similarly when detecting potential clusters. However, for different levels of spatial aggregation, the potential clusters identified were different. As the number of regions used in the analysis increased, the total area identified in the cluster decreased. In general, potential clusters were identified in the central and northern parts of Manitoba. Overall, it is crucial to identify the appropriate number of regions to study spatial patterns of disease as it directly affects the results and consequently the conclusions. Additional investigation through future work is needed to determine which scenario of spatial aggregation is best.

Dimensions

Altmetric

PlumX Metrics

Downloads

Download data is not yet available.

Citations

Supporting Agencies

Children’s Hospital Research Institute of Manitoba, Natural Sciences and Engineering Research Council of Canada

How to Cite

Torabi, M., & Galloway, K. (2018). Examining the impact of the number of regions used in cluster detection methods: An application to childhood asthma visits to a hospital in Manitoba, Canada. Geospatial Health, 13(2). https://doi.org/10.4081/gh.2018.696

List of Cited By :

Crossref logo