Among men living in the United States, prostate cancer is the second leading cause of cancer death, and, excluding skin cancers, it is the cancer diagnosed most frequently. While incidence and mortality rates have been declining, the American Cancer Society estimated that there were 220,800 men diagnosed with prostate cancer and more than 27,500 prostate cancer deaths in 2015. Various patient-level and community-level factors have been shown to influence the differential patterns of diagnosis, care, and outcomes for men with prostate cancer. Detailed information regarding the utilization of health services by prostate cancer patients, particularly those with higher propensity for health services use, could be used to inform efforts intended to improve the coordination and delivery of care to work towards the elimination of disparities. The purpose of the study is to facilitate a better understanding of the determinants of health services utilization by older males with prostate cancer in the United States by examining the relative influence and interaction effects of factors characterizing individual patients and their county of residence. Andersen's behavioral model of health services utilization is used as a framework to guide this study. A cross-sectional design is used to analyze administrative claims data from the 2008 Medicare Provider Analysis Review (MEDPAR) file (n=5,754). County-level data from Area Health Resources File (ARHF) are merged to include the community and contextual characteristics. American Hospital Association (AHA) annual survey data are also used to examine the importance of hospital attributes in a subset analysis (n=555). A two-stage approach is used for analyzing the data. First, several social and demographic variables are included in automatic interaction detector (AID) analysis to identify relatively homogenous subgroups of patients with similar service utilization patterns for emergency room visits and hospital length of stay. Second, regression analysis is performed in the full dataset including all patients, and in each subgroup to determine the amount of variance explained by predictor variables categorized as predisposing, enabling, and need-for-care factors. Hierarchical logistic regression is performed to analyze the variability in emergency room use, and hierarchical multiple regression is performed to analyze the variability in hospital length of stay. The results show that the need-for-care factors are dominant predictors of service use. However, the relative importance of the predictor variables varies by subgroups of prostate cancer patients identified in the initial AID analysis. The findings lend some support of the use of an integrated approach to examine the personal and social determinants of health services utilization by prostate cancer patients enrolled in the U.S. Medicare program. The theoretical framework and analytic approach employed in this study make it possible to obtain an in-depth understanding of the influential factors associated with emergency room use and length of stay for all-cause hospitalizations, which can be used to inform future research and efforts aimed at developing targeted interventions to improve the coordinated care and to reduce health disparities among Medicare beneficiaries with prostate cancer.


If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at STARS@ucf.edu

Graduation Date





Wan, Thomas T. H.


Doctor of Philosophy (Ph.D.)


College of Health and Public Affairs

Degree Program

Public Affairs; Public Administration









Release Date

August 2019

Length of Campus-only Access

3 years

Access Status

Doctoral Dissertation (Open Access)