The disparities in the distribution of goods and services, and hazards and opportunities across space are increasing, underscoring the growing connection between place and health. Although ample evidence confirms that living in an economically disadvantaged neighborhood is associated with adverse health outcomes, the reliance on cross-sectional data and inadequate attention to two main sources of bias make causal inferences problematic. Residents tend to sort themselves into different types of neighborhoods based on a multitude of characteristics. Not accounting for all characteristics that are correlated to both the outcome and neighborhood context would likely lead to over-estimations of neighborhood effects. Because regression models cannot possibly account for all relevant factors, the strong possibility of unobserved heterogeneity make neighborhood effect studies open to criticisms of omitted variable bias. Yet, at the same time, neighborhood effect studies are also just as likely to be susceptible to bias due to over- adjustment. Many factors that are controlled for in neighborhood effect models, such as educational attainment, income, and employment, may arguably have been influenced by past neighborhood context. Adjusting for these factors eliminate possible critical pathways through which neighborhoods affect health, likely yielding overly conservative estimates of neighborhood effects. These two sources of bias, working in opposing directions, have plagued extant neighborhood-health research;consequently, results from current research yield tenuous and ambiguous inferences. This proposed project will use novel analytical methods and longitudinal data from an existing observational study to address the two major limitations described above and recover causal estimates of neighborhood poverty on self-rated health and mortality. We will 1) use marginal structural modeling to appropriately adjust for covariates that are simultaneously confounders as well as mediators and 2) conduct a sensitivity analysis to determine the robustness of the neighborhood effect findings to unobserved heterogeneity. Applying this combined methodology to neighborhood-health research has the potential to significantly advance our knowledge of the relationship between place and health, yielding far reaching policy implications.
Relevance Robust findings of a causal connection between residential context and health can help health policymakers judge the extent and magnitude of neighborhood impacts on health and guide public health policies. Evidence that the social and structural environment influences life-chances, and ultimately health outcomes, suggests that health policy, traditionally targeted at the individual level with little regard to neighborhood context, should consider underlying constraints or opportunities present in the residential environment in designing and implementing the most effective and efficient public health policies.
Do, D Phuong; Zheng, Cheng (2017) A marginal structural modeling strategy investigating short and long-term exposure to neighborhood poverty on BMI among U.S. black and white adults. Health Place 46:201-209 |
Do, D Phuong; Wang, Lu; Elliott, Michael R (2013) Investigating the relationship between neighborhood poverty and mortality risk: a marginal structural modeling approach. Soc Sci Med 91:58-66 |