Regional disparities in health and mortality in the U.S. have been observed repeatedly, but little attempt has been made to explain them. At best, anecdotal explanations are usually offered. For example, poorer health among southerners is often attributed to diet without any empirical support. Extant literature suffers from several additional shortcomings. First, many studies focus on a single health outcome, like stroke mortality, thereby underestimating the full extent of regional variation in health. Second, many studies measure region coarsely. Often, only one region is contrasted against all others. This approach also leads to underestimation of the full range of regional variation in health and hinders our ability to understand the precise mechanisms that account for it, because within-region cultural and structural heterogeneity is extensive. Third, studies of regional disparities have generally failed to take a life course perspective, instead treating them as existing in a temporal vacuum. The proposed research will address these shortcomings, first by adopting a life course perspective. The life course perspective recognizes that neither region of residence, nor health, nor the relationship between them, is static at the individual level across age. Furthermore, regional characteristics and the distribution of health outcomes also vary across sociohistoric time, implying that the relationship between region and health may differ across birth cohorts. The life course perspective therefore provides a more comprehensive and detailed lens through which to begin to explain regional differences in health. Given this perspective, the proposed research will establish the full extent of regional disparities in health using a variety of longitudinal statistical methods applied to at least three nationally-representative, large sample data sets: the General Social Survey, the Health and Retirement Study, and the National Health Epidemiologic Follow-up Surveys. These data will be augmented via the collection of region-year contextual variables like physician density, climate, etc. Collectively, these surveys contain a wide variety of health measures, including self-rated health, physical functioning, depressive symptoms, mortality, and diabetes, as well as refined measures of region (i.e., the nine-category Census measure). Importantly, these three surveys also contain at least one measure of region of residence in early life (birth and adolescence), which, from a life course perspective, is useful in helping differentiate the role of early life socialization into regional culture from the role of structural characteristics of an individual's current region of residence in influencing health. In addition, this early life region measure, as well as the use of longitudinal methods, will enable the investigation of the extent to which health influences regional mobility, an issue (i.e., endogeneity) commonly ignored in research. Basic descriptive methods, typical regression models, multistate life table methods for both panel and cross-sectional data, and hierarchical growth models, including autoregressive latent trajectory models, will be used to flesh out the extent of regional differences in health as well as the mechanisms that account for them.
Regional differences in health and mortality in the US have been observed in numerous studies but have been left largely unexplained. Understanding the mechanisms that account for regional differences is important for determining how health disparities can be reduced. Our proposed project aims to determine the full extent of regional differences in health and to explain them (1) by using more refined measures of region and a broader array of health outcomes than used in previous research, and (2) by employing a life course perspective and methods to differentiate with greater precision and accuracy the pathways via which region may affect health.
|Berchick, Edward R; Lynch, Scott M (2017) Regional Variation in the Predictive Validity of Self-Rated Health for Mortality. SSM Popul Health 3:275-282|