Document Type

Capstone Experience

Graduation Date


Degree Name

Master of Public Health



First Committee Member

Chistopher Wichman

Second Committee Member

Jianghu Dong

Third Committee Member

Kate Grafel


Health insurance companies have a goal of improving population health for their members (people who the company insures). As a health insurance company, [Company] has ample data on the health of its members that can be utilized to improve the health, and by extension lives of the people they insure. Although [Company] does not deliver care, they communicate with their members and physicians to identify ways to improve the health of the member. Of specific interest are members with known chronic conditions who are receiving suboptimal care, as well as members who have undiagnosed chronic conditions. Two chronic conditions were chosen to demonstrate the feasibility of statistical modeling to help identify areas for improvement: diabetes and the combination of asthma and or chronic obstructive pulmonary disease (COPD). For each condition, two models using the generalized estimating equation approach were created for a total of four models. The first model classifies members with known chronic conditions who are receiving suboptimal care as identified by a hospital visit caused by their condition. The second model finds members with a chronic condition that is yet undiagnosed. After completing these two models, it is apparent that in all cases the status of a preventative visit and each additional chronic condition a member has will increase the odds of an event. It appears drug count and various distinct conditions have differing impacts on outcome dependent upon the chronic condition of interest to the model.