Date of Award

12-14-2015

Document Type

Open Access Thesis

Department

Epidemiology and Biostatistics

First Advisor

Jan M. Eberth

Abstract

AIM: The primary aim of this study was to evaluate the geographic variation in emergency department (ED) use in South Carolina using geographical information systems (GIS) and to examine the effects of both individual- and neighborhood-level characteristics on frequent ED use among patients with diabetes. METHODS: ED discharge data for 2013 was geocoded based on patients’ residential ZIP code using GIS. Frequent ED use among patients with diabetes was defined as three or more ED visits between January 1, 2013 and December 31, 2013. The boundaries for each neighborhood were defined by the U.S. Census Bureau ZIP Code Tabulation Areas (ZCTAs) and the demographic and socioeconomic characteristics describing each ZCTA were obtained from the 2013 American Community Survey. Population standardized density of ED patients with diabetes as well as the average number of ED visits per patient with diabetes were calculated for each ZCTA and analyzed for spatial patterns and non-spatial correlations with neighborhood-level determinants. The relationships between individual- and neighborhood-level variables with frequent ED use were assessed using random-intercept multilevel modeling. RESULTS: A total of 350 out of 423 ZCTAs were included in this analysis, with a sample size of 91,461 ED patients with diabetes who accumulated over 166,905 ED visits in South Carolina during 2013. The standardized density of ED patients with diabetes as well as the average number of ED visits per patient with diabetes demonstrated spatial clustering to certain geographic locations within South Carolina. Indicators of low neighborhood socioeconomic status and family structure were significantly correlated with a higher density of patients with diabetes and an elevated number of ED visits per patient with diabetes. In multilevel analyses, patients with diabetes who were younger, African American females, or Medicaid/Medicaid beneficiaries were more likely to be frequent ED users. At the neighborhood-level, measures of material deprivation and rural/urban status of a neighborhood demonstrated little to no effect on the propensity toward using the ED in this population. CONCLUSION: Exposure to certain neighborhood-level characteristics may increase or decrease an individual’s dependence upon the ED for routine care. However, after accounting for individual-level characteristics via multilevel modeling, neighborhood-level measures of material deprivation and rural/urban status did not account for much of the unexplained neighborhood heterogeneity. Whereas, individual-level measures of age, sex, race, and primary source of payment remained significant predictors of frequent ED use. The spatial clustering of ED patients with diabetes and elevated ED visits per patient with diabetes to certain geographic locations in South Carolina indicates that future research should spatially explore this relationship in order to understand the behavior process leading to ED utilization in this high risk population.

Included in

Epidemiology Commons

Share

COinS