| dc.description.abstract | The main goal of this dissertation research is to explore the impacts of rural hospital acquisition on patients’ access to tertiary care and treatment patterns, as well as to predict the medical care utilization provided by rural hospitals by applying a Bayesian hierarchical modeling approach.
In the first two studies, I target cross-market rural hospital acquisition between a tertiary hospital (i.e., acquiring hospital) and a local community hospital (i.e., acquired hospital) and investigate impacts of hospital acquisition on local patients’ access to tertiary care and treatment patterns using Difference-in-Difference approach with Texas Inpatient Public Use Data Files (PUDF). Patients’ access to tertiary care is measured on two levels: (1) on ZIP code (i.e., hospital market) level, the acquiring hospital’s market share of patients before and after the acquisition; (2) on discharge (i.e., patient) level, whether a patient would be admitted by an acquiring hospital. Patients’ treatment patterns are measured as: (1) on ZIP code level, the proportion of patients receiving an interventional treatment and the acquiring hospital’s market share of the interventional treatment; (2) on discharge level, whether a patient would receive an interventional treatment and whether an interventional treatment would be performed at an acquiring hospital. I find that the impacts of rural hospital acquisition are different by the market competition status, various types of care, and patients’ characteristics. When there is no competing hospital in the same market as the acquired hospital, the impacts on access to tertiary care are positive for inpatient newborn and cardiovascular care. The impacts are different by patients’ expected payer source and severity of illness. A similar pattern is observed in the investigation of impacts on treatment patterns.
In the third study, I apply a Bayesian hierarchical modeling approach to predict the newborn delivery utilization at small hospitals in rural areas using the PUDF data. The results show that the Bayesian approach can provide a more accurate predication on the medical service utilization than a maximum likelihood approach, indicating that the Bayesian approach application might support a rational allocation of limited health care resources for rural hospitals. | en |
