Date of Award

Spring 2017

Degree Name

Master of Arts



First Advisor

Dr. Kathleen M. Baker

Second Advisor

Dr. Benjamin Ofori-Amoah

Third Advisor

Dr. Amy B. Curtis

Fourth Advisor

Dr. Rajib Paul


Diabetes, spatial analysis, regression, Geographic Information System, model selection

Access Setting

Masters Thesis-Abstract Only

Restricted to Campus until



In the U.S, diabetes has become one of the major health concerns. In like manner, health insurance coverage has become vital to the health needs of individuals in U.S. Adults having elevated glucose levels are recommended to receive glycated hemoglobin (HbA1c) testing to determine the average blood sugar concentrations. These screenings help in reducing costs related to diabetes complications and hospitalization. Differences in insurance coverage can have a significant impact on which patient obtain recommended tests.

The study analyzes secondary data for a period of three years (2011-2013) for three different health plans being Medicaid, Blue Care Network (BCN) and Blue Cross Blue Shield (BCBS) for the state Michigan. Spatial and statistical methods were used to evaluate and ascertain the best regression model for count data and the association between county specific health and socioeconomic factors and insurance plans associated with the HbA1c testing. The study finds the negative binomial model is best in predicting count health data. Also, urban-rural interface and type of insurance plan are key in understanding patterns and frequency of diabetes service utilization. The BCBS plan has more people taking the required HbA1c test compared to the BCN and Medicaid. Overall, persons with private insurance appear to be conscious and that the required HbA1c testing. Across all plans, the analysis indicates that the southern part of Michigan needs more attention. Thus interventions should be focused on this area. Further, health and socioeconomic factors determine the rate and frequency of the HbA1c testing.