Spatial Analysis of Lung Cancer Mortality in the American West to Improve Allocation of Medical Resources

Abstract

Over 80% of lung cancer incidence in the USA has been linked with smoking, yet causes of lung cancer mortality (LCM) are more complex and have been linked with a range of cultural, environmental, economic and health, lifestyle variables. These all vary spatially yet spatial studies of lung cancer are rare. This paper investigates spatial patterns in county-level LCM and the factors related to it in the western US which has markedly lower rates of LCM than the eastern US. Two variables, not previously investigated, that could be partially responsible for lower rates in the west are included in the analysis. These are elevation and membership of the Church of Jesus Christ of Latter-day Saints (LDS). Analysis involved aspatial and spatial regression, geographically weighted regression (GWR) and univariate and bi-variate local Moran’s I (LMI) cluster analysis. Regression showed that the explanatory power of covariates varies with the area or region studied although the LDS population and elevation, were important both nationally and within the west. GWR analysis showed how correlations between the dependent and independent variables, and regression coefficients, vary spatially within the western region. This showed broad and smooth trends in the non-stationarity of correlations and coefficients with generally different behavior in the coastal compared to mountain west states. Bivariate LMI analysis was helpful in identifying small clusters of problem counties. These included counties in WY and CO that produce oil and gas with large transient male populations that smoke, counties in CA associated with severe air pollution problems and poor rural counties in several states that have high rates of unemployment and lack of medical facilities. This improved understanding of spatial patterns could be used so that screening/educational efforts and location of the medical facilities can be improved to target the locations and population groups most at risk.