Abstract P296: Investigation Of Geographic Disparities Of Pre-diabetes And Diabetes And Associated County-level Risk Factors In Florida, 2016

Abstract

Background: Diabetes and its complications represent a significant public health burden in the United States, with evidence of geographic disparities. Identifying these disparities and their determinants is useful for guiding control programs. Therefore, this study investigated geographic disparities of pre-diabetes and diabetes prevalence in Florida in 2016, and identified predictors of the observed spatial patterns. Additionally, we investigated changes in geographic distribution of the two conditions between 2013 and 2016. Methods: The 2013 and 2016 Behavioral Risk Factor Surveillance System data were obtained from the Florida Department of Health. Flexible scan statistics were used to identify significant high prevalence geographic clusters. Global ordinary least squares regression and local Poisson geographically weighted generalized linear models were used to investigate predictors of the identified spatial patterns. Counties with significant changes in prevalence of the two conditions between 2013 and 2016 were identified using tests for equality of proportions adjusted for multiple comparisons using Simes method. Results: The state-wide diabetes prevalence was 11.2% in 2013, and 11.8% in 2016. Statistically significant ( p ≤0.05) increases in prevalence were identified in 73% (49/67) of the counties. Similarly, the state-wide prevalence of pre-diabetes was 7.1% in 2013 and 9.2% in 2016 with 76% (51/67) of the counties reporting statistically significant increases. Significant local hotspots were identified for both conditions. Predictors of county-level diabetes prevalence were: proportion of the obese population, number of physicians per 1000 persons, proportion of the population living below the poverty level, and proportion of the population with arthritis. Predictors of pre-diabetes prevalence included proportion of the population with arthritis and proportion of the population that identified as non-Hispanic black. There was evidence of geographical variability of all regression coefficients for both the pre-diabetes and diabetes models indicating that the strength of association of the relationships between the predictors and outcomes varied by geographic area. Conclusions: Geographic disparities of both conditions continue to exist in Florida. Moreover, there was a state-wide increase in the burden of both conditions between 2013 and 2016. The fact that the strength of association of the relationships between the predictors and outcomes varied across the counties implies that some predictors may be more important in some counties than others. These findings imply that local models provide useful information to guide public health decision-making and resource allocation. Identifying high-risk geographic areas and location-specific determinants of chronic disease prevalence should be used to inform targeted intervention programs.