Epidemiology of diabetes and related mortality: early screening, socioecological determinants, and the value of prevention

The focus of my dissertation is prediction, prevention, and economic valuation of type 2 diabetes. I studied individual level type 2 diabetes risk factors, spatial spillover effect of diabetes-related mortality ecological risk factors, and individuals’ loss of well-being due to diabetes. In the first essay, titled “Diabetes Risk Prediction: Multivariate Nonlinear Interaction Approach,” I argue that the success in preventing or delaying the incidence of type 2 diabetes and subsequent complications depend on the early detection of undiagnosed cases and identifying people at high-risk. However, early detection of type 2 diabetes is seldom feasible because the symptoms show up late, and screening the entire population is very costly. Individuals who are prone (e.g., due to family history) to developing type 2 diabetes and those with undiagnosed diabetes need to be targeted for early screening. Thus, it is imperative to continue designing assessment mechanisms that help to identify individuals at high-risk based on simple, non-invasive, inexpensive, and routine clinical measurements. In this paper, I build a model that helps to predict type 2 diabetes with readily available, inexpensive, non-invasive, and easy-to-collect information. National Health and Nutrition Examination Survey (NHANES) data is analyzed to build this risk model. A non-parametric regression method, Multivariate Adaptive Regression Splines (MARS), is used to allow for interactions and non-linearity in the model. A risk prediction model using the MARS approach achieved a performance level of 87% accuracy with area under receiver operating character curve (AUROC) of 0.86, which is higher than similar models based on invasive and non-invasive measurements. Moreover, this model requires few measurements and limited information that may be obtained in settings such as community-based chronic disease prevention programs and workplace well-being programs. Therefore, this risk prediction model can be translated into a usable risk-scoring tool in community chronic prevention and employee wellness programs. The second essay, titled “Spatial Spillover Effect from Socio-Ecological Determinants of Diabetes-Related Mortality in the US,” explores the spatial spillover effect from socio-ecological risk factors that are associated with type 2 diabetes-related mortality. I studied the spatial spillover effect of change in socioeconomic gradients (education, employment, and household income), retail food environments, and access to health-care on diabetes-related mortality rates (DRMR) across the United States. To examine mortality clusters and factors associated with the clusters and spatial spillover effect, seven-year aggregates of multiple-cause mortality data from CDC WONDER compressed mortality database was merged with several sources of county-level data. The results show that high DRMR cluster counties are located throughout the Southern Plains, Southeastern, and Appalachian regions. High DRMR clusters are characterized by lower socioeconomic status, high density of fast food restaurants, lack of access to grocery stores, high proportion of African Americans, and low physical activity. Moreover, the impacts from change in socioeconomic gradients and the retail food environment in a particular county spill over to neighboring counties. The implication is that improvement in socioeconomic status and access to healthy food would significantly reduce DRMR in contiguous US counties. The third essay, titled “What is the Value in Diabetes Prevention? A Subjective Well-Being Valuation Approach,” uses loss of well-being due to diabetes to quantify the monetary value of diabetes prevention in the US population. In this paper, I argue that the current preference-based health valuation approach is not appropriate for prevention-based programs valuation because they do not capture the social and economic value that an individual puts on a health condition. I utilize a recently developed subjective well-being valuation approach to quantify the monetary value of loss in well-being due to diabetes in the US population. This approach assumes that individuals derive overall life satisfaction from well-being, which is a function of health and income. Health, in turn, is produced by the combined input of an individual’s behaviors and medical technology. Thus, a marginal trade-off between health and income is used to derive the monetary value of health. The Panel Study of Income Dynamics (PSID) data was utilized for this study. The result shows that the monetary value for diabetes prevention is about $37,000, which is less than the current implicit threshold for program implementation. The resulting monetary value will help to quantify the societal value of diabetes prevention, which can be used to estimate the benefit side of the cost-benefit analysis