"In one region of the US, poverty and lower education outcomes are more predictive of higher diabetes prevalence, and in other regions, physical inactivity and obesity are more predictive," said lead author J. Aaron Hipp, assistant professor from Washington University's Brown School in the US.

The study suggests that approaches to combating the disease should be localised.

Hipp and co-author Nishesh Chalise analysed county data from the US Census Bureau and the Centres for Disease Control and Prevention to find attributes associated with diabetes. They discovered that results varied by region.

Poverty levels and inactivity were associated with diabetes, but only in some areas. The percentage of the population cycling or walking to work correlated with lower prevalence of diabetes in most counties, but not in some rural areas of Minnesota, North Dakota and South Dakota.

"Given this clustering of predictors of diabetes prevalence, and knowing the effect of the predictors we used in our study, counties, states and regions should be able to better target the most common predictors of diabetes in their more local area," Hipp added.

Latest News from Lifestyle News Desk