Multiscale Analysis of Cancer Service Areas in the United States

Keywords: Cancer care markets, cancer service areas (CSAs), spatially constrained Leiden (ScLeiden) method, multiscale, spatial network community detection
Abstract Type: Paper Abstract

Abstract

The purpose of delineating Cancer Service Areas (CSAs) is to define a reliable unit of analysis, more meaningful than geopolitical units, for examining geographic variations of the cancer care markets using geographic information systems (GIS). This study aims to provide a multiscale analysis of the U.S. cancer care markets using the spatial network created from the Medicare claims of cancer-directed surgery, chemotherapy, and radiation. The CSAs are delineated by a scale-flexible and GIS-automated network community detection algorithm so that the patient flows are maximized within CSAs and minimized between them. The multiscale CSAs include those comparable in size to 4 census regions, 9 divisions, 50 states, and 39 global optimal CSAs. The CSAs are more effective in capturing the U.S. cancer care markets because of higher localization index, lower cross-border utilization, and shorter travel time. The first two comparisons reveal that only a few regions or divisions are representative of the underlying cancer care markets. The last two comparisons find that among the 39 CSAs, 54% CSAs comprise multiple states anchored by cities near inner state borders, 28% are single-state CSAs, and 18% are sub-state CSAs. Their (in)consistencies across state borders or within each state shed new light on where the intervention of cancer care delivery or the adjustment of cancer care costs are needed to meet the challenges in the U.S. cancer care system. The findings could guide stakeholders to target public health policies for more effective coordination of cancer care in improving outcomes and reducing unnecessary costs.