Diabetes Miletus (DM) is one of the major health problems in the United States. Despite all efforts made to combat this disease, its incidence and prevalence are steadily increasing. One of the common and serious side effects of treatment among people with diabetes is hypoglycemia (HG), where the level of blood glucose falls below the optimum level. Episodes of HG vary in their severity. Nevertheless, many require medical assistance and are usually associated with higher utilization of healthcare resources such as frequent emergency department visits and physician visits. Additionally, patients who experience HG frequently have poor outcomes such as higher rates for morbidities and mortality. Although many studies have been conducted to explore the risk factors associated with HG as well as others that looked into the level of healthcare utilization and outcomes among patients with HG, most of these studies failed to establish a theoretical foundation and integrate a comprehensive list of personal risk factors. Therefore, this study aimed to employ Andersen's health Behavior Model of health care utilization (BM) as a framework to examine the problems of HG. This holistic approach facilitates enumerating predictors and examining differential risks of the predisposing (P), enabling (E) and need-for-care (N) factors influencing HG and their effects on utilization (U) and outcomes (O). The population derived from the national inpatient sample of the Healthcare Cost and Utilization Project (HCUP) database and included all non-pregnant adult diabetic patients admitted to hospitals' Emergency Departments (EDs) with a diagnosis of HG from 2012-2014. Based on the BM framework, different factors influencing HG utilization and outcome were grouped under the P, E, or N component. Utilization was measured by patients' length of stay (LoS) in the hospital and the total charges incurred for the stay. Outcome was assessed based on the severity ranging from mortality (the worst), severe complications, mild complications, to no complications (the best). Structural Equation Modeling (SEM) followed by Decision Tree Regression (DTREG) were performed. SEM helped in testing multiple hypotheses developed in the study as well as exploring the direct and indirect impact of different risk factors on utilization and outcome. The results of the analysis show that N is the most influential component of predictors of U and O. This is parallel to what was repeatedly found in different studies that employed the BM. Regarding the other two components, P was found to have some effect on O, while E influences the total charge. Interaction effects of predictors were noted between some components, which indicate the indirect effect of these components on U and O. Subsequently, DTREG analysis was conducted to further explore the probability of the different predictor variables on LoS, total charge, and outcome. Results of this study revealed that the presence of renal disease and DM complications among HG patients play a key role in predicting U and O. Furthermore, age, socio-economic status (SES), and the geographical location of the patients were also found to be vital factors in determining the variability in U and O among HG patients. In conclusion, findings of this study lend support to the use of the BM approach to health services use and outcomes and provide some practical applications for healthcare providers in terms of using the predictive model for targeting patient subgroups (HG patients) for interventions among diabetic patients. Moreover, policy implications, particularly related to the Central Florida area, for decision makers regarding how to approach the growing problem of DM can be drawn from the study results.
Wan, Thomas T. H.
Doctor of Philosophy (Ph.D.)
College of Health and Public Affairs
Public Affairs; Health Services Management and Research
Length of Campus-only Access
Doctoral Dissertation (Open Access)
Kattan, Waleed, "Factors Influencing Hypoglycemia Care Utilization and Outcomes Among Adult Diabetic Patients Admitted to Hospitals: A Predictive Model" (2017). Electronic Theses and Dissertations, 2004-2019. 5416.