Graduation Date

Spring 5-4-2019

Document Type


Degree Name

Doctor of Philosophy (PhD)


Health Services Research, Administration, and Policy

First Advisor

Li-Wu Chen

Second Advisor

Jeffrey Harrison


Background: It has become a national priority to reduce the high health care expenditure in the United States while improving the quality of care. Hospital care is taking up one-third of the healthcare spending, and services offered in hospitals are costly compared to others. Only one-twentieth of the patients with high-needs account for about half of the health care spending. They consuming a high level of hospital services if their conditions are not well-managed in the outpatient settings. Therefore, it is important to examine the effectiveness of the approaches that have the potentials to reduce costly care utilization through improvements in the quality of care. This dissertation thesis focused on examining the effects of three approaches to reduce hospital utilization. The three approaches include the patient-centered medical homes (PCMH), better continuity of care (COC), and the early use of inpatient palliative consultation (IPC) at the end of life.

Methods: Andersen’s Behavioral model of health care utilization was used to guide the modeling process of the three individual studies. The first study used data from the Medical Expenditure Panel Survey Household Component (MEPS-HC). Respondents who reported having a usual source of care other than the emergency department (ED) were included, and they were classified into three levels of PCMH groups by their baseline-year care features from 11 selected items. The outcomes were the second-year hospital admissions and ED visits due to the ambulatory care sensitive conditions (ACSCs). Logistic regressions that accounted for survey weights were used. The second study was conducted among a nationally representative Taiwan Population who were admitted for the first time for the five conditions. The outcomes are the numbers of all-cause and condition-specific hospitalizations during the follow-up year after discharge, and the primary explanatory variable was the outpatient COC. Multivariable generalized estimation equation models with a negative binomial distribution and log link were used. The third study used Nebraska Hospital Discharge Data linked with death certificates to identify the inpatient services received by the Nebraska Decedents due to the top six causes of death. The use of IPC was classified by the time receiving it as early use and late use, and the comparison group was the decedents who never encountered IPC. The outcomes were end-of-life events including hospice discharge, place of death, intensive care utilization, life-sustaining treatment, length of stay and total inpatient charges. Mixed-effect logistic regressions, logistic regression, negative binomial regression, and generalized linear model with log link and gamma distribution were used for those outcomes respectively.

Results: The highest level of PCMH primary care was associated with lower risks of having admissions and ED visits due to ACSCs. However, individual attributes of PCMH did not have the same effects. The patients with better COC have significantly fewer all-cause hospitalizations for all the conditions. The COC only worked in patients with ACSC conditions in reducing the condition-specific hospitalizations. The early use of IPC was associated with lower likelihoods of dying in the hospitals, receiving intensive care and the life-sustaining treatment. The use of IPC at either the early or late time was associated with higher odds of being discharged to hospice care, and less length of stay in the inpatient settings and less total inpatient charges.

Conclusion: Approaches such as PCMH, improving continuity of care and the early use of palliative care are promising in reducing the costly hospital services and improving the quality of care. These approaches are replicable to any value-based programs for cost-reduction, quality improvement, and improving population health outcomes.