Search

Background : The dichotomization or categorization of rural-urban codes, as nominal variables, is a prevailing paradigm in cancer disparity studies. The paradigm represents continuous rural-urban transition as discrete groups, which results in a loss of ordering information and landscape continuum, and thus may contribute to mixed findings in the literature. Few studies have examined the validity of using rural-urban codes as continuous variables in the same analysis. Methods : We geocoded cancer cases in north central Florida between 2005 and 2010 collected by Florida Cancer Data System. Using a linear hierarchical model, we regressed the occurrence of late stage cancer (including breast, colorectal, hematological, lung, and prostate cancer) on the rural-urban codes as continuous variables. To validate, the results were compared to those from using a truly continuous rurality data of the same study region. Results : In term of associations with late-stage cancer risk, the regression analysis showed that the use of rural-urban codes as continuous variables produces consistent outcomes with those from the truly continuous rurality for all types of cancer. Particularly, the rural-urban codes at the census tract level yield the closest estimation and are recommended to use when the continuous rurality data is not available. Conclusions : Methodologically, it is valid to treat rural-urban codes directly as continuous variables in cancer studies, in addition to converting them into categories. This proposed continuous-variable method offers researchers more flexibility in their choice of analytic methods and preserves the information in the ordering. It can better inform how cancer risk varies, degree by degree, over a finer spectrum of rural-urban landscape.

Effects of urban/rural residence on late-stage cancer have long been explored, but remained controversial. Spatial granularity of rural definition, temporal change of rurality, and local variability of such effects may contribute to inconsistent findings, but they have not been fully addressed. We proposed a spatially resolved and temporally comparable rurality index and a geographically weighted regression approach to re-examine this question. Taking Florida as an example, our analyses show that rural effects on late-stage cancer vary dramatically over locations (600-m cells). The odds ratios range from 0.9 to 1.10, and imply that one degree of rurality can increase/decrease local risk of late-stage cancer by up to 10%. Our study is an early attempt to explore local effects of rurality on cancer at a fine spatial scale, and reveals interesting patterns hidden by global multi-level analysis. The new framework and findings can better inform precision interventions to mitigate cancer disparities., (Copyright © 2018 The Authors. Published by Elsevier Ltd.. All rights reserved.)

Population demand, health service supply, and the linkages between them (e.g., transport infrastructure) are important factors that determine spatial accessibility to healthcare at a place. These three factors vary differently over time and location, leading to temporal changes and spatial disparities in access to healthcare. Few analytic methods have been developed to measure local impacts of these factors on healthcare accessibility over time, which are essential to alleviating health disparities and evaluating intervention programs. We propose a spatially explicit analytic framework to measure local factor impacts over time by adopting a chain substitution method from economics. The analysis is illustrated by a case study of spatial accessibility to physicians in Florida, USA, from 1990 to 2010. For each census block group, the results show the impact of local population change, physician relocation, and road-network expansion on the loss and gain of healthcare accessibility over time. The leading impact factor are identified for each census block group through comparison, and spatial clusters of factor impacts are discovered. To the literature of healthcare accessibility, this article presents a promising start of factor impact analysis and offers new perspectives in exploring spatial processes underlying people's access to healthcare., (Copyright © 2018 Elsevier Ltd. All rights reserved.)