Doctor of Philosophy (Ph.D.)
Degree Granting Department
Industrial and Management Systems Engineering
José L. Zayas-Castro, Ph.D.
Jay Wolfson, Dr.P.H.
Robert Frisina, Ph.D.
Mingyang Li, Ph.D.
Ankit Shah, Ph.D.
Cesarean delivery, Machine learning, Pediatric readmission, Policy analysis, Predictive modeling
Over the last two decades, the United States has spent almost twice as much per person in healthcare compared to most other wealthy countries. However, this higher spending has not necessarily transformed into improved quality of care; According to World Health Organization reports, the US now ranks 39th for child health and wellbeing and worst in maternal care among developed nations. In terms of proportion of preventable hospital visits, low-risk cesarean sections, and avoidable maternal morbidity/death, the U.S. is among the highest compared with the peer nations. The prevalence of these adverse outcomes in pediatric and obstetric care is particularly disproportionately high among Medicaid beneficiaries in comparison to privately insured patients, mainly driven by persistent disparities in access to care and care experiences. Consequently, Medicaid expenditure for these groups has been straining federal and state budgets in the last decades, and a substantial increase is expected in the future. In the US, nearly half of obstetric and more than a third of pediatric healthcare is provided through the Medicaid program, and the Medicaid system continues to face substantial challenges in improving care quality and reducing cost in what is now a major policy concern. The key challenges in improving the quality of child and maternal health services provided through Medicaid are (1) how to enhance understanding about the causes and implication of pediatric care fragmentation, higher preventable hospital visits and cesarean rates, and (2) how to better design decision support for Medicaid patients that considers all major stakeholders, which can reduce adverse outcomes, improve health in the vulnerable population and consequently, saves money for the American people.
The objectives of this dissertation, therefore, were to generate new knowledge regarding care fragmentation and disparities in pediatric and maternal health and to develop improved, data-informed decision support that aims to reduce the adverse outcomes associated with Medicaid settings. Using the Florida State and national claims databases, fragmentation of pediatric care was explored in the context of index vs non-index readmission, then associated risk factors were identified, and finally impact of this difference in destination effect on readmission outcomes were explored. Furthermore, after illustrating novel geographical and racial disparities in the fragmented context of pediatric care and the adverse implications of non-index readmission, ways of improving pediatric readmission prediction were explored that could aid both managed care programs and hospitals in designing comprehensive interventions that target children who are at high risk for readmission. More specifically, two innovative decision support approaches were proposed to enhance the prediction of pediatric readmission as compared with existing approaches. First, a novel early risk predictive model was proposed at the time of hospital admission that improves the high-risk patient selection process for hospitals. In the second approach, a cohort-specific readmission model was proposed that achieved higher discrimination when compared with traditional all-cause readmission models. In addition, an innovative framework of preventable ED visits and revisit prediction models at three patient-provider interaction timepoints under Medicaid managed care settings was proposed in this dissertation. This model has practical applicability for managed care organizations and can help improve the patient selection process for intervention planning, particularly for services targeting the social determinants of children’s health and wellbeing.
For improving maternal care quality, the causes of the persistently high interstate variations in cesarean rates were investigated and their implications on financial and adverse health outcomes were analyzed. Finally, the impact of the Florida Statewide Medicaid Managed Care (SMMC) programs on pediatric and maternal care outcomes were estimated with a focus on reducing racial and ethnic disparities. After the SMMC implementation, there was a substantial reduction in several pediatric and maternal care outcomes and associated disparities. The findings of this study could help state policymakers understand the current performance of existing SMMC programs in reducing care disparities as well as facilitate the design of better policies and managed care contracts.
In summary, through the development of these six studies, this dissertation comprehensively provides novel insights and introduces innovative decision support approaches considering all major Medicaid stakeholders, which can be used to better design Medicaid pediatric and maternal care delivery systems.
Scholar Commons Citation
Symum, Hasan, "Data-Informed Decision Support to Improve Pediatric and Maternal Care Quality Under Medicaid Managed Care Settings" (2021). Graduate Theses and Dissertations.