Doctor of Philosophy (PhD)


Department of Geography and Anthropology

Document Type



Delineating a meaningful and reliable geographic unit pertaining to cancer care is essential in examining geographic variations in cancer care for better analysis, management, and planning. Public health researchers have devoted great efforts to defining various service areas, which; however, lack scientific rigor or are unrepresentative of the highly specialized cancer care markets, and their methods have become obsolete in the era of big data and high geo-computation. This study develops the “Spatially Constrained Leiden (ScLeiden)” method, a network community detection algorithm in Geographic Information Systems (GIS), and applies it to delineate a series of coherent cancer service areas (CSAs) in the U.S.

Using the national Medicare enrollment and claims, this study creates a spatial network of cancer cares across the U.S. where each network edge represents the service volume of cancer-directed surgery, chemotherapy, and radiation between ZIP codes. It then develops and automates the ScLeiden method in an easy-to-use GIS toolkit. The method explicitly maximizes the service volumes within defined areas and accounts for spatial contiguity and other constraints. Its derived service areas outperform existing areas in terms of self-containment and size comparability.

The method is applied to define multiscale CSAs in comparable numbers to census regions, divisions, and states. The results provide valuable guides for stakeholders to more effectively coordinate cancer care and advance efforts in cancer control and prevention across and within states. The National Cancer Institute mandates its designated cancer centers to identify catchment areas for assessing cancer burden and needs. To assist this effort, this study refines the ScLeiden method to delineate nationwide CSAs anchored by major cancer centers, identified by the Association of American Cancer Institutes. The result can be used as reliable units to evaluate cancer care in areas within versus outside the catchment areas of these designated cancer centers, and inform more evidence-based policy.

The work will have a major impact on the research and stakeholder communities engaged in evaluating geographic variations in health care delivery and outcomes. The toolkit and CSAs will be widely disseminated, thus empower them to define any units pertaining to their specific health care systems and needs.



Committee Chair

Wang, Fahui