If you had bothered to click on the link to the original study, you would have seen this in the second paragraph (the "Methods" section):

"Linear regressions of BMI and logistic regressions of overweight and obesity include controls for individual-level age and neighborhood-level racial/ethnic composition, median age of residents, and median family income."