County-level Cigarette Smoking Prevalence Estimates

We applied previously described small area models to estimate the prevalence of cigarette smoking for US counties [21 (link)-23 (link)]. In brief, we constructed a family of logistic hierarchical mixed effects regression models for each outcome, stratified by sex. These models incorporate spatial and temporal smoothing and a series of county- and state-level covariates to improve predictions for all counties, including those with limited data available in a given year from the BRFSS. More details on the regression models and the county- and state-level data sources incorporated in the models can be found in Additional files 1 and 2. These models allowed us to generate annual estimates of total and daily cigarette smoking prevalence for male and female adults (age 18 and older) in all US counties and county equivalents. All estimates were age-standardized following the age structure of the 2000 census [24 ]. The uncertainty of the prevalence estimates was assessed using simulation methods [25 (link)].