Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Health Outcomes and Policy Research


Health Policy

Research Advisor

James E. Bailey, M.D., M.P.H.


Cameron M. Kaplan, Ph.D. Robert J. Nolly, M.S., D.Ph. George E. Relyea, M.S. Jim Y. Wan, Ph.D.


BACKGROUND: Medicaid coverage among adults is often characterized by discontinuity – loss of Medicaid coverage, and churning, or entering and exiting Medicaid – over short durations. Little is known about the impact of having discontinuous Medicaid coverage on access to care, preventive care, pharmacotherapy, primary care, and hospitalizations, among non-elderly adults with cardiovascular disease (CVD) or high-risk conditions.

OBJECTIVES: This dissertation employed a three empirical research papers approach to pursue the following aims: (1) characterize the adult subpopulations with CVD or conditions placing them at high risk for CVD who lack continuous Medicaid coverage, and examine the characteristics associated with Medicaid discontinuity, (2) examine associations between Medicaid discontinuity, medication adherence and medication utilization, and (3) examine the associations between Medicaid discontinuity, access to care, preventive care, primary care visits, and hospitalizations.

METHODS: This was a retrospective comparative analysis of the 2002–2011 Medical Expenditure Panel Survey employing a repeated cross-sectional study design. Study sample included adults aged 18–64 years diagnosed with ≥1 CVD (defined as acute myocardial infarction, coronary artery disease, congestive heart failure, peripheral and visceral atherosclerosis, or stroke) or high-risk conditions for CVD (defined as hypertension, lipid disorders, diabetes, or chronic kidney disease) who reported having Medicaid coverage any time during survey year. Individuals having continuous, full-year Medicaid coverage (Continuous Medicaid) were compared to those with(Discontinuous–Uninsured), and separately to those with(Discontinuous–Insured). Associations between Medicaid discontinuity, access to care, and preventive care were estimated using multivariate logistic regression. Medication adherence, measured as medication possession ratio (MPR) with adequate adherence being considered at MPR>0.8, was estimated using multivariate logistic regression. Medication utilization, measured as the number of all-cause, and disease-specific prescription drug fills, was estimated using multivariate negative binomial regression. Four health care services utilization outcomes – inpatient, emergency room (ER), hospital outpatient, and office-based physician visits – measured as both, number of all-cause and number of disease-specific visits, were estimated using either zero-inflated negative binomial regression or negative binomial regression depending on the distribution of the outcome of interest.

RESULTS: Overall, 31.8% of adults with CVD or high-risk conditions for CVD hadcoverage, majority of whom (23.5%) belonged to the Discontinuous– Uninsured group. Of those who had Medicaid at the beginning of the year, only 21.9% of the Discontinuous–Uninsured, and 8% of the Discontinuous–Insured still had Medicaid by the year end. Male gender, minority race/ethnicity, receiving disability benefits or participating in a federal assistance program, Medicaid managed care enrollment, and diagnosis of respiratory illnesses were the characteristics associated with lower odds of Medicaid discontinuity, whereas being married, residing in the South, having higher income, or education, being employed, and having fair to poor perceived health status were associated with higher odds of Medicaid discontinuity. Overall adherence to commonly prescribed therapeutic medication classes, measured as average MPR, was not significantly different between the Continuous Medicaid and the two discontinuous coverage groups, whereas examination of class-specific adherence yielded mixed results. Discontinuous Medicaid coverage was associated with significantly lower allcause and disease-specific prescription drug utilization among both the discontinuous Medicaid groups. Medicaid discontinuity was associated with poor access to care, and higher diseasespecific inpatient and ER hospitalizations among both the discontinuous Medicaid coverage groups. Additionally, among the Discontinuous–Uninsured, Medicaid discontinuity was associated with lower odds of routine medical checkup, lower all-cause primary care office visits, and higher disease-specific hospital outpatient visits.

CONCLUSION: Among non-elderly adults with CVD or high-risk conditions, for CVD having discontinuous Medicaid coverage was found to be associated with poor access to care and preventive care, poor adherence to certain medication classes and lower utilization of prescription medications, higher hospitalizations for CVD or associated conditions, and lower primary care office visits. Disruptions in and loss of Medicaid coverage among adults with CVD or high-risk conditions may lead to negative health outcomes due to the disruptions in continuity of care and inability to appropriately manage these disease conditions. This research provides strong support for implementation of policies to stabilize Medicaid coverage and reduce Medicaid discontinuity among individuals with CVD or high-risk conditions for CVD, such as the 12 month continuous Medicaid eligibility provisions currently in place for low-income children. Such policies will greatly improve some of the adverse access to care, preventive care, pharmacotherapy, and medical care outcomes observed in this study. Vulnerable populations with chronic, debilitating conditions may benefit from such policies to a greater extent compared to the overall low-income adult population. Simultaneously, reenrollment and outreach strategies may need to be more efficiently implemented to ensure individuals who are eligible for Medicaid, continue to remain enrolled in Medicaid. Such enabling strategies employed by Medicaid managed care organizations may be adopted by State Medicaid agencies to ensure greater continuity in Medicaid coverage for low-income vulnerable populations with chronic and debilitating diseases.



Available for download on Saturday, December 31, 2016