Measurement matters in rural-urban health disparities: Regional variation in obesity-rurality associations in older adult based on choice of rurality scale

Although rural-urban health disparities are well documented and constitute an important cause of health inequalities across the US, there remains a lack of a unifying definition of “rural" in the US.  To extend work previously presented, we assess internal consistency among five common measures of rurality used in epidemiological research in US counties by Census region and division, profile metropolitan areas, and examine how the associations between rurality and obesity depend upon the rurality measurement chosen as a methodological example.  We abstracted five measures of rurality from the 2010 US Census and USDA to characterize US counties: percent urban, rural-urban continuum codes, urban influence codes, population density quartile, and an Index of Relative Rurality.  County-level weighted obesity prevalence was abstracted from the 2010 Behavioral Risk Factor Surveillance System.  In the Northeast, there were strong rank correlations among the five measures (rho 0.756-0.964), while the South had the weakest correlations (0.453-0.941).  Associations between obesity prevalence and rurality varied substantially by region.  In the Northeast, increasing rurality was significantly associated with a higher prevalence of obesity for all five measures.  However, in the South, increasing urbanicity was significantly associated with a decrease in obesity prevalence for population density only (p=0.007) and not associated with the other four measures. These findings highlight internal inconsistencies among rurality measures, regional differences in consistency among the measures, and the need to develop meaningful and universal rural-urban measures for epidemiological research to determine the specific aspects of rurality and urbanicity that drive rural-urban health disparities.