Abstract

Lung and bronchus cancer is one of the leading causes of death in the United States. Much research has found that this cancer is closely associated with poverty, and incidences of this cancer do not appear to be distributed homogeneously across space. Studies have investigated spatial variations of cancer incidence at various spatial resolutions, such as countries, cities, counties, and census units, ignoring their urban hierarchy context. In general, an urban area can be categorized as a global, national, or regional place according to the standard urban hierarchy typology, and cancer patterns may differ across these urban hierarchy classes. The purpose of this chapter is twofold. First, it empirically investigates whether or not lung and bronchus cancer rates have different patterns across three urban hierarchy classes and poverty levels using cancer incidence in six MSAs in Florida for the time period 2006 to 2010. Second, it examines whether or not location errors associated with cancer incidence have an impact on the relationship of cancer rates to urban hierarchy typology and poverty, because location errors for cancer incidence often are embedded intentionally to preserve confidentiality, or unintentionally owing to data processing steps such as geocoding. This chapter utilizes eigenvector spatial filtering to account for spatial autocorrelation in regression models. Results indicate that the geographic distribution of cancer is different in terms of poverty and urban hierarchy typology, and confirm that location errors have an impact on analysis outcomes. Because this study essentially is the first of its kind, it contributes to the formulation of expectations about direction and magnitude of these effects for future studies. Presently, the only expectation is that significant differences exist.