An analysis of potentially undiagnosed diabetes and diabetes-related complications at local, regional and national scales
Delmerico, Alan M.
MetadataShow full item record
This research is divided into three major research areas on the related topics of undiagnosed diabetes and diabetes-related complications, a local-level analysis of potentially undiagnosed diabetes, a regional analysis of hospitalizations for diabetes-related complications, and a national-level analysis of diabetes-related complications using individual-level data. The local-level analysis assesses potentially undiagnosed diabetes among a general population sample of individuals ages 15–45 in Erie County, NY. Residuals from a model of diabetes diagnosis were examined for spatial autocorrelation to determine if individuals with elevated risk for diabetes but who lack a diagnosis are clustered in space. The findings from this section indicate a strong spatial variation in individual-level risk for diabetes diagnosis. This spatial heterogeneity in Erie County, NY indicates a cluster of potentially underdiagnosed diabetes in the urbanized area of the City of Buffalo and its first-ring suburbs. The methods employed offer a leading example for the use of spatial health data related to diabetes and other chronic disease diagnoses to identify co-located “surprise” events (e.g. individuals who have the risks but not the diagnosis). While the data used here are specific to Erie County, NY, these analyses can serve as a template for utilization of other spatial health data for the spatial targeting of community-based disease screening. The regional-level analysis assesses neighborhood-level factors that drive hospitalizations for diabetes-related complications. These outcomes are predominantly preventable with proper diabetic management and care and are related to neighborhood-level sociodemographic characteristics such as poverty and education. This analysis provides a better understanding of the spatial variation in these outcomes at a spatial scale that is meaningful, with potentially actionable interventions by community health promotion actors. Specifically, these results suggest that social determinants of health provide good explanation for neighborhood-level differences in hospitalizations for diabetes-related complications overall in the diverse built environments of two definitions of the Western New York region. For both regional definitions, there are more hospitalizations for diabetes-related complications than expected from higher concentrations of African Americans, lower concentrations of Hispanics, lower educational achievement, higher poverty rates, higher domicile mobility, higher rates of single parent households, and lower rates of housing vacancy. Social determinants of health and disparities in general, appear to be major drivers of hospitalizations for diabetes-related complications. The results from this component of the analysis suggest that the explanation of these outcomes from social determinants of health is uneven over space, and more specifically built environment type, with these factors explaining the vast majority of variance in excess hospitalization in urban areas. Their more limited explanation in suburban and rural areas suggests that there are additional, un-modeled factors that are important in determining these outcomes in these areas. Additionally, the persistent spatial autocorrelation in the urban-specific model residuals suggests that the process underlying hospitalizations from diabetes-related complications is spatially heterogeneous in these areas. These results can help guide broader public health interventions and strategies that are informed by social determinants of health and can be aligned across a continuum of care to bring appropriate levels of service to locations that need them and provide linkages between the prevention and treatment systems. The national-level analysis evaluates structural and contextual components of diabetes-related complication development with more specific diabetes-related data while controlling for personal diabetic management information that is presently not readily available at other spatial scales. The results from this component of the analysis indicate that differences in complications between counties and between states are minimal. However, these outcomes do not appear to be wholly explained by individual-level factors or by the contextual, group-level factors of urban-rural continuum type, rates of uninsured, and the prevalence of diabetes within the general population. This suggests the possibility of unobserved factors, possibly group-level social determinants of health in this case, affecting between-group differences in outcomes. The removal of spatial dependence in the eye complications model residuals with the addition of group-level covariates suggests that these factors are important in explaining the spatial pattern of these outcomes. It is interesting however that this is not observed for the foot complications models. The results presented across the three scales of this research suggest the importance of using spatial and other data to drive the local health response to diabetes and other chronic behavioral health problems. While this research explores crucial topics in diabetes such as the driving social factors affecting diabetic outcomes, it pays particular attention to the spatial aspects of these health issues in order to answer questions of where these issues occur. Additionally, it provides insights into how and where interventions can be implemented to reduce the overall burden of these health outcomes on our health care system and our society as a whole. (Abstract shortened by UMI.)