November - 20 - 2009
Model-based estimates may help identify areas with populations at high risk of diabetes and obesity, according to a report by a team of researchers with the Centers for Disease Control (CDC). The researchers used Bayesian multilevel models to estimate the prevalence of diabetes and obesity at the local level in all 3,141 U.S. counties in 2007. The counties with a diabetes prevalence of 10.6 percent or more were primarily located in a belt that extended from the Mississippi River to the coastal Carolinas and Appalachians. Nearly 75 percent of the counties in Alabama, Georgia, Louisiana, the Carolinas, and the Appalachians were in the top quintile for diabetes prevalence, as were more than 60 percent of the counties in West Virginia and Tennessee combined. High prevalence was also noted in isolated counties, such as tribal lands in Montana, the Dakotas, the Southwest, and eastern Oklahoma. The median estimate of obesity prevalence was 28.4 percent, and counties with the highest prevalence were primarily in the South, western Appalachians, and coastal Carolinas. The data had some limitations, including the use of self-reporting and the exclusion of undiagnosed diabetes. Local health officials may use the results of this model to help develop programs for reducing diabetes and obesity.