Comparison of small-area analysis techniques for estimating prevalence by race on the county level

Background: The Behavioral Risk Factor Surveillance System (BRFSS) is commonly used for estimating the prevalence of chronic disease. One major limitation of the BRFSS is that valid estimates can only be obtained for state and larger geographic regions. There is limited health data available on the county level and thus many have used small area analysis techniques to estimate the prevalence of disease on the county level using BRFSS data.

Methods: In this study the validity and precision of five small area analysis techniques for estimating the prevalence of three chronic diseases, asthma, diabetes, and hypertension by race on the county level were compared. Prevalence estimates produced by direct estimation, synthetic estimation, spatial smoothing, temporal estimation, and regression were compared to gold standard estimates obtained through local data collection using six discrepancy statistics, Pearson and Spearman correlation coefficients, mean square error, mean absolute difference, mean relative absolute difference, and rank statistics.

Results: The regression method produced estimates of the prevalence of chronic disease by race on the county level that had the smallest discrepancies with the gold standard estimates for a large number of counties.

Conclusions: When applying small-area analysis techniques to obtain county level prevalence estimates of chronic disease by race using a single year of BRFSS data, regression is the preferable method. A beta version of a web-based portal was developed to deliver county level estimates to non-technical audiences.