Methodologic changes in the behavioral risk factor surveillance system in 2011 and potential effects on prevalence estimates